STEM CELLS AND CORNEAL EPITHELIAL REGENERATION

FRIEDRICH E. KRUSE Heidelberg, Germany

SUMMARY movement of cells. The vertical movement can be docu­ Self-renewing tissues such as the corneal epithelium con­ mented experimentally by the chase of previously labelled tain stem cells which represent the proliferative reserve. basal cells and might be due to the proliferative pressure in Studies of cellular differentiation and proliferation sug­ the basal cell layer.5 The horizontal movement of corneal gest that corneal epithelial stem cells are localised exclus­ epithelial cells from the periphery to the centre was ively in the basal limbal epithelium. Although regulatory observed after experimental corneal epithelial wound­ factors for the amplification of corneal stem cells are ing.6.7 Similarly in rabbits, the centripetal replacement of unknown, serum factors such as retinoic acid might corneal epithelium after corneal grafts which originates in induce differentiation of stem cells to transient ampli­ the donor epithelium has been shown to start in the periph­ fying cells which are responsible for cell amplification. ery of the graft.R Tracing of peripherally located ink par­ These cells are regulated by various polypeptide growth ticles in the murine epithelium has proved that the factors and extracellular calcium. Loss or malfunction of centripetal movement of corneal epithelial cells also takes stem cells does not permit maintenance or regeneration of place under physiological circumstances in normal animal the corneal epithelial mass but leads to conjunctival­ corneas.9 Several observations indicate that the centripetal isation of the corneal surface. Clinically, several ocular movement exists in human corneas as well. First, corneal surface disorders such as chemical burns can cause lim­ erosions which do not include the limbal epithelium heal bal damage and consecutive limbal insufficiency. Treat­ in a centripetal fashion.lo Second, small subepithelial ment for these disorders is available only by cysts which develop between the sutures of corneal grafts transplantation of healthy stem cells, which can be per­ I I move towards the centre when the sutures are removed. formed as both autograft and allograft. Third, a centripetal movement of epithelial cells under physiological conditions was observed by specular 12 PRINCIPLES OF CORNEAL EPITHELIAL microscopy. REGENERATION The proliferation of basal cells as well as horizontal and Corneal epithelium is subject to a constant process of cell vertical cell movements have been summarised in kinetic renewal and regeneration. Cells in the uppermost layer of models which describe the maintenance of the corneal epi­ the corneal epithelium are continuously desquamated thelial mass. lUi While these mechanisms of the mainten­ from the surface and must be replaced by cell prolifer­ ance of corneal epithelium are generally accepted, the ation. The exclusive localisation of dividing cells in the question of the origin of corneal epithelial cell prolifer­ basal layer of the corneal epithelium suggests that pro­ ation has sparked a considerable controversy. Two oppos­ liferation is limited to basal cells. 1-4 Only cells which are ing theories exist, one of which claims that the origin of in contact with the basement membrane have the ability corneal epithelium is derived from the adj acent conjunc­ for mitotic cell division while cells which are displaced tiva by conjunctival transdifferentiation while the other into the suprabasal layers become post-mitotic and lose claims that the origin of corneal epithelial proliferation their capability for cell division.s depends on corneal stem cells in the limbal basal The kinetics of the maintenance of the corneal epithelial epithelium. mass is characterised by a vertical as well as a horizontal CONJUNCTIVAL Correspondence to; Dr med. Friedrich Eduard Kruse. Augenklinik TRANSDIFFERENTIATION der Universitat Heidelberg, 1m Neuenheimer Feld 400. 0-69120 Heidelberg, Germany. The first theory is based on early studies of corneal wound

Eye (1994) 8,170-183 © 1994 Royal College of Ophthalmologists STEM CELLS OF THE CORNEAL EPITHELIUM 17 1 healing in humans which observed an ingrowth of con­ longer exposure to n-heptanol in combination with mech­ junctival epithelium onto the denuded cornea following anical scraping led to trans differentiation in only 32% of large epithelial wounds extending beyond the limbus.15.16 the eyes.20-27 In the remaining eyes conjunctival epithe­ Further observations and experimental studies showed lium did not transdifferentiate but was supported by neo­ that in the absence of vascular ingrowth conjunctival epi­ vascularisation. These findings suggest that the extent of thelium on the corneal surface loses its conjunctival phe­ removal of the corneal epithelium is responsible for the notype and becomes cornea-like.16-18 This transformation incidence of transdifferentiation or persistent conjunctiv­ of conjunctival epithelium to a corneal epithelium was alisation. This observation draws attention to the role of described by the term 'conjunctival transdifferentiation'. limbal epithelium in the maintenance and restoration of This phenomenon was investigated in numerous animal corneal epithelium. It is tempting to speculate that com­ studieslY-22 and led to the assumption that normal corneal plete removal of both corneal and limbal epithelium epithelium is maintained by the surrounding conjunctival induces irreversible conjunctivalisation. On the contrary, epithelium.13 incomplete removal of the limbal epithelium may allow However, two lines of evidence have cast doubt upon the reconstruction of the original epithelium aftera certain the existence of true conjunctival transdifferentiation. period which is needed for recovery of the damaged epi­ First, the histological and ultrastructural appearance of thelium and which can falsely be interpreted as the time transdifferentiated conjunctival epithelium differs from sequence of conjunctival transdifferentiation. that of genuine corneal epithelium, for example with To prove this hypothesis we investigated the interaction respect to the presence of goblet cells.23 Second, the phe­ between the duration of corneal exposure to n-heptanol notype of the epithelium after transdifferentiation shows and the extent of corneal and limbal epithelial removal by distinct differences from the phenotype of genuine corneal means of a histological survey.33 The results indicate that epithelium with respect to its metabolism as well as its exposure of the corneal epithelium to n-heptanol with composition of proteins and keratins.2�-26 These data indi­ mechanical scraping results in complete removal of the cate that conjunctival transdifferentiation does not repre­ epithelium when the agent is applied for more than 60 sent the true conversion of a differentiated conjunctival seconds. In contrast, the limbal epithelium was much phenotype into a differentiated corneal phenotype but more resistant to this treatment. Exposure of the limbal rather describes an environmental modulation of the con­ epithelium to n-heptanol resulted in incomplete removal junctival epithelium. This notion is supported by the of the basal layer even when the duration of treatment was observation that conjunctival transdifferentiation can be extended to 180 seconds. These results indicate that the inhibited by vascularisation.16.17 In contrast, occlusion of treatment used in the aforementioned investigations to the vessels in conjunctivalised epithelium induces con­ remove the corneal epithelium completely (n-heptanol junctival transdifferentiation.27 treatment of less than 120 seconds) most probably also These observations suggest that the blood vessels resulted in incomplete removal of the limbal basal supply conjunctival epithelium on the corneal surface epithelium. with substances which prevent transdifferentiation or To investigate further whether the remaining basal lim­ which are - in other words - important for the mainten­ bal epithelium retained its proliferative capacity despite ance of the conjunctival phenotype. Numerous studies by, the treatment we conducted tissue cultures of corneo­ for example, Tseng and coworkers have demonstrated that scleral specimens from eyes which had been treated with retinoic acid is of great importance for the differentiation n-heptanol for 60 and 90 seconds following mechanical of goblet cells under physiological conditions and that ret­ scraping of the epithelium." In these cultures we observed inoic acid can prevent conjunctival trans differentia­ a continuous outgrowth from the limbal epithelium onto tion.28-31 It is therefore tempting to speculate that a the denuded stroma. Immunohistochemical staining with localised deficiency of retinoic acid (and other unidenti­ various antibodies showed that the outgrowth was of lim­ fied factors) results in a loss of goblet cells and a modu­ bal derivation. These results indicate that the remaining lation of conjunctival epithelium into a cornea-like basal epithelium retains its proliferative capacity and epithelium. therefore can reconstitute an epithelial phenotype after A closer analysis of the experiments which were per­ removal of the corneal epithelium as observed in the afore­ formed to induce conjunctival transdif ferentiation suggest mentioned transdifferentiation experiments. In the light of another explanation for transdifferentiation in addition to these results experimentally induced trans differentiation environmental modulation. The vast majority of the could be explained by the following sequence: First, experiments investigating transdifferentiation used a removal of the epithelium by n-heptanol leads to incom­ model in which chemical and mechanical removal of cor­ plete removal of the basal limbal epithelium. Second, con­ neal and limbal epithelium causes the conjunctival epi­ junctival epithelium overgrows the limbal basal thelium to move onto the cornea.32 These experiments, epithelium and leads to conjunctivalisation of the corneal however, resulted in great variability concerning the inci­ epithelium. Third, the remaining limbal epithelium recov­ dence of the resulting transdifferentiation. Application of ers and replaces the conjunctival epithelium with corneal a chemical agent (n-heptanol) for a short period caused epithelium which appears as conjunctival transdifferentia­ transdifferentiation in 86% of the treated eyes while tion. These findings highlight the role of the basal limbal 172 F. E. KRUSE epithelium for the maintenance of the corneal epithelial cells is based mainly on the observation that only the lim­ mass under physiological conditions and its importance bal basal epithelium contains cells which exhibit two of for corneal regeneration after epithelial defects. This leads the aforementioned characteristics of stem cells. First, the to the second theory regarding the origin of the corneal Iimbal basal epithelium contains the least differentiated epithelium which claims that the origin of the corneal epi­ cells of the corneal epithelium. Second, the limbal basal thelium lies in corneal stem cells located in the limbal epithelium contains cells which exhibit the proliferative basal epithelium. characteristics of stem cells. Davanger and Evansen34 were the first to speculate that corneal epithelium derives from the limbal pallisades of The Basal Limbal Epithelium Contains the Least Vogt. This hypothesis was based on the observation that D(fferentiated Cells of the Corneal Epithelium pigmented epithelial migration lines which occur in The differential expression of keratins first allowed the heavily pigmented eyes migrate from the limbus towards separation of cell populations within the corneal epithe­ the centre of the cornea. Ten years later Schermer et al. 35 lium according to their level of differentiation. Keratins conducted a survey of the cellular differentiation of cor­ are intermediate-type filaments which naturally occur as neal and limbal epithelium with respect to a certain class pairs consisting of an acidic keratin with its matching of intermediate filaments, i.e. keratins. The most impor­ basic keratin. Among the more than 20 keratins which tant conclusion of this study was the hypothesis that cor­ have been classified accordingto their isoelectric focuss­ neal epithelium originates from the limbus and more ing point by Moll and coworkers.42 some indicate a high precisely that the limbal basal epithelium contains the level of differentiation while others are found mostly in stem cells of the corneal epithelium. less differentiated cells.42.43 The development of specific antibodies by Sun and EPITHELIAL STEM CELLS coworkers3S.44,45 first allowed identification of the location The presence of stem cells is postulated in all self-renew­ of the 64 kDa keratin K3 which indicates a cornea-specific ing tissues, where they serve as the reserve for cell renewal type of differentiation. It was observed that keratin K3 and cell proliferation.36-38 Epithelial stem cells share the exists in the suprabasal epithelium of the limbus and the following characteristics which have also been postulated entire corneal epithelium but is expressed in neither the for the stem cells of the corneal epithelium: 39..l0First , stem limbal basal epithelium nor the adjacent bulbar conjunc­ cells have a long life span which might be equivalent to the tiva. This observation led to the hypothesis that the limbal life of the organism which harbours them. Second, they basal epithelium lacks a differentiated cornea-type pheno­ have an almost unlimited potential for (colo go genic) cell type and therefore contains the least differentiated cells of division. Third, stem cells are slow cycling, which indi­ the epithelium, i.e. stem cells.35 Further studies have cates a low mitotic activity. Fourth, stem cells are the least shown that the basal limbal epithelium also lacks the differentiated cells in the tissue and lack markers which expression of the second half of the corneal-specifickera­ indicate greater differentiation. Fifth, stem cells can be tin, pair, i.e. K12, which is expressed in the suprabasal induced on demand, by certain factors to differentiate into limbal epithelium and in the entire corneal epithelium.-l6,47 transient amplifying cells. In contrast to stem cells tran­ In summary these findings suggest that the basal limbal sient amplifying cells are characterised by a high mitotic epithelium is less differentiated than the suprabasal limbal rate but a limited proliferative capacity. At a higher level epithelium and the entire corneal epithelium with respect of differentiation than stem cells, transient amplifying to the expression of cornea-specific keratins. cells embark on a pathway leading to further differentia­ Studies of the expression of other keratins which indi­ tion and ultimately to cell death. After a high but limited cate a relatively undifferentiated phenotype show further number of cell divisions transient amplifying cells further evidence for the low level of differentiation oflimbal basal differentiate to post-mitotic cells which have lost the epithelium. In humans keratin K19 can always be found in capacity for cell division. These post-mitotic cells then limbal basal epithelium and sometimes in the basal epi­ become terminally differentiated and die after a certain thelium of the periphery. In contrast, the suprabasal limbal time. and corneal epithelium do not express this keratin in most In contrast to haematopoietic stem cells, which have cases.48 Since K19 is expressed in the entire limbal and been positively identifiedby antibodies,4! the existence of corneal epithelium of the human fetus its presence in the stem cells has not been positively proven in any of the basal limbal epithelium of the adult might indicate remaining self-renewing tissues such as the epithelium of embryogenetically young cells, i.e. stem cells. On the con­ the skin or the ocular surface. However, a wide body of trary, K19-positive cells in the peripheral corneal epi­ indirect evidence suggests the presence of stem cells in the thelium mightbe the result of centripetal cell movement, corneal epithelium as well. since keratin K 19 has also been associated with regenerat­ ing basal epithelium. LOCATION OF CORNEAL EPITHELIAL Recent studies of the co-expression of various classes STEM CELLS IN THE BASAL LIMBAL of intermediate filaments have identifieda subset of limbal EPITHELIUM and peripheral corneal basal cells which do not express Evidence for the limbal location of corneal epithelial stem keratin K3 but co-express keratin K19 and vimentin.49 STEM CELLS OF THE CORNEAL EPITHELIUM 173

Since this cell type was almost exclusively present in the the number of labelled cells after exposure to TPA. It basal epithelium of the superior limbal circumference, it therefore seems likely that not only limbal stem cells but was speculated that these cells might either represent a early transient amplifying cells which are also present in morphologically distinct subset of stem cells or migrating corneal epithelium were stimulated by the tumour transient amplifying cells.49 promoter. Several attempts have been made to develop bio­ In tissue cultures TPA has been used also to differen­ chemical or immunological markers which are specificfor tiate between stem and transient amplifying cells. Treat­ limbal stem cells. Zieske and coworkers50 have generated ment of epithelial cell cultures with TPA induces terminal a monoclonal antibody against a protein with the molecu­ differentiation in the vast majority of the cells, which lar weight of 50 000 which is exclusively expressed in therefore stop proliferating. These cells represent transient limbal basal epithelium. Although it was initially hoped amplifying cells.55-57 Only a fraction of the cultured cells that this protein represents a marker specific for limbal can retain their undifferentiated phenotype and proliferate stem cells, further studies have shown that it represents the despite the presence ofTPA. These TPA-resistant cells are glycoloytic enzyme alpha-enolase which also occurs in the least differentiated cells in the epithelium, i.e. the stem other tissues.51 Its significance and function in limbal basal cells. 55-57 epithelial cells are currently being investigated. In order to prove further the existence of limbal stem All of the investigations described have identified the cells on the basis of their response to TPA in tissue culture limbal basal epithelium as a group of cells with properties we used a serum-free clonal growth assay which allows differing from the remaining corneal epithelium. How­ the comparison of single cells of limbal and corneal epi­ ever, these investigations were not able to differentiate thelium.58 In this assay TPA-resistant colonies were between different subpopulations within the limbal basal observed in cultures not only from limbal but also from epithelium. Nevertheless it is most likely that the limbal peripheral and central cornealepith elium. 59 Both the abso­ basal epithelium consists not only of stem cells but also of lute number of TPA-resistant colonies and the percentage transient amplifying cells. An approach to differentiating of resistant colonies relative to the total number of colo­ between these cell populations within the limbal basal epi­ nies in control medium was significantly higher in cultures thelium is by investigating their proliferative behaviour. from limbal epithelium than from corneal epithelium. 59 Although most TPA-resistant colonies were observed in The Basal Limbal Epithelium Contains Cells with cultures from limbal epithelium, the presence of TPA­ the Proliferative Behaviour of Stem Cells resistant colonies in corneal epithelial cultures as well Short-term labelling of cells with agents that identify does not seem to be consistent with the concept of the mitotic divisions (such as tritiated thymidine, eH] TdR) exclusive location of corneal epithelial stem cells in the allows the identification of fast cycling transient ampli­ limbus. It might however, be possible that TPA-resistant fying cells whereas prolonged administration of the label colonies derive not only from limbal stem cells but also enables the identification of slow cycling stem cells.52,53 from a second cell type which is present in both limbal and Using these labelling characteristics Cotsarelis and co­ corneal epithelium, i.e. early transient amplifying cells. workers54 were the firstto differentiatebetween stem cells This interpretation is suggested by the observation that the and transient amplifying cells in the basal limbal epi­ pattern of differentiation and proliferative behaviour of thelium. After short-term labelling they observed proli­ TPA-resistant colonies in cultures from both limbal and ferating cells in both the limbal and peripheral corneal corneal epithelium was almost identical. Furthermore, basal epithelium, indicating the presence of transient earlier studies showed that the tissue culture environment amplifying cells. They then applied [3H] TdR for 14 days used in these studies promotes epithelial differentiation and extended the observation period to 4 weeks. Under and that it is probably not possible to culture true stem these conditions only a very few labelled cells could be cells on plastic substratum in serum-free medium.59 Taken identifiedin the basal limbal epithelium (and none in the together these findings suggest the existence of a pool of cornea). The authors interpreted this observation by sug­ very early transient amplifying cells with stem cell charac­ gesting that the cycling time of the limbal stem cells is teristics (i.e. TPA resistance) in both limbal and corneal longer than 2 weeks.54 epithelium. The existence of such cells would also explain In order to improve the chances of detecting stem cells, the clinical observation that the corneal epithelium can be these quiescent progenitor cells can be exposed to sub­ maintained for a long period even when the limbal basal stances which increase their level of differentiation and epithelium is largely missing. recruit them into a state of proliferation. Such agents To establish further the existence of limbal stem cells which induce cellular differentiation are tumour promo­ Tseng and coworkers used the antimetabolite 5-ftuoroura­ ters or antimetabolites. Topical application of the tumour cil (5-FU) which also allows the identification of early promoter TPA (12-0-tetradecanoylphorbol-13-acetate) progenitor cells because of its differentiation-inducing resulted in a preferential stimulation of the proliferation of effect. Both in vitro and in vivo application of 5-FU limbal epithelial cells in comparison to corneal epithelial showed that the limbal epithelium contained significantly cells.54 However, not only the limbal but also the periph­ more cells with stem cell characteristics than the corneal eral corneal epithelium showed a significant increase in epithelium.60 These investigations confirmedthe presence 174 F. E. KRUSE of stem cells and a pool of early progenitor cells which is mental evidence. First, the original corneal phenotype can present in both limbal and corneal epithelium. not be maintained or reconstituted in the absence of stem In summary these studies suggest that the limbal epi­ cell-containing limbal epithelium. Second, the original thelium contains a population of slow cycling cells which phenotype of corneal epithelium can be reconstituted by display the proliferative characteristics of stem cells. surgical transplantation of limbal stem cells. These cells are also resistant to the induction of differ­ entiation. In addition, limbal, peripheral and central cor­ Wound Healing in the Absence of Corneal Stem neal epithelia seem to contain a population of very early Cells transient amplifying cells which display some of the The regeneration and maintenance of the corneal epi­ characteristics of stem cells. thelium in response to partial or total removal of corneal Both the investigations of differentiation and the inves­ stem cells at the limbus was investigated by a series of tigations of proliferation of Iimbal and corneal epithelium experiments performed on rabbits by Tseng and co­ can be summarised by a model of the location of corneal workers.6I.63 In the presence of corneal stem cells within epithelial stem cells as shown in Fig. 1. The stem cells are an uninjured limbal epithelium corneal epithelium regen­ located exclusively in the Iimbal basal epithelium. The erates despite repeated small central wounds, even if the occurrence of transient amplifying cells in limited to the total corneal epithelium is removed. basal layer of limbal and corneal epithelia. Post-mitotic When corneal stem cells are depleted by removing two and terminally differentiated cells make up the superficial thirds of the limbal epithelium, the remaining stem and layers oflimbal and corneal epithelia. This model describ­ transient amplifying cells can maintain the corneal epi­ ing the location of corneal epithelial stem cells in limbal thelium under physiological circumstances. This state of basal epithelium is further supported by an analysis of the depletion of corneal stem cells has been named partial Iim­ kinetic maintenance of the corneal epithelial mass. This bal deficiency.61 When the central corneal epithelium is mathematical model confirms that the corneal epithelium removed in eyes with partial limbal deficiency, rapid can be maintained by cellular proliferation originating wound healing was observed in 75% of the eyes even after from limbal stem cells without contribution of the adj a­ two consecutive central woundings.61 This indicates that cent conjunctiva.14 the remaining stem and transient amplifying cells not only maintain the epithelial mass but also regenerate the central IMPORTANCE OF CORNEAL STEM CELLS corneal epithelium in the majority of the cases. On the FOR THE REGENERATION OF THE other hand 25% of the eyes in which two thirds of the lim­ CORNEAL EPITHELIUM bal epithelium was missing developed vascularisation and The location of corneal epithelial stem cells in Iimbal delayed wound healing, indicating that eyes with partial basal epithelium suggests that the regeneration of corneal limbal deficiency are at risk of developing wound healing epithelium is highly dependent on the integrity of the lim­ problems after the additional loss of central transient bus. This hypothesis is supported by two lines of experi- amplifying cells. When most of the transient amplifying LIMBUS CORNEA cells were removed in eyes with partial Iimbal deficiency by creating a large defect of the central and peripheral cor­ neal epithelium, all of the eyes investigated showed delayed wound healing, vascularisation and expression of a conjunctival phenotype.61 These findings indicate that the remaining stem cells could not generate enough tran­ sient amplifying cells to reconstitute the corneal epithe­ lium and suggest a loss of the limbal barrier against the ingrowth of conjunctival epithelium on the surface of the cornea. These findings also support the importance of transient amplifying cells for the maintenance of the cor­ neal epithelium. In particular the early transient ampli­ fying cells which partially display stem cell characteristics and exist in both limbal and corneal epithelium59 maintain Terminally ? differentiated and regenerate the corneal epithelium even in the absence Postmitotic cells of a large portion of limbal stem cells. f'("cells '; When all the limbal stem cells are surgically removed -+ Transie�� resulting in a total limbal deficiency, the remaining tran­ eStemcells amplifying sient amplifying cells maintain the corneal epithelial mass cells under physiological conditions in two thirds of the ani­ Model of the limbal location of com eo stem cells. Stem Fig. 1. I mals while the remaining develop mild vascularisation.62 cells (black) are exclusively located in the basal limbal epi­ thelium. Transient amplifying cells occur ollly ill the basal epi­ After two consecutive central to mid-peripheral corneal thelia of' limbus and cornea. Post-mitotic and terminally woundings 75% of the corneas showed vascularisation differentiated cells make up the superficial layers. and ingrowth of conjunctival epithelium.62 These results STEM CELLS OF THE CORNEAL EPITHELIUM 175 further support the importance of Iimbal stem cells but transplantation significantly reduced the area of neovas­ also indicate that even in the complete absence of stem cularisation and allowed restoration of the corneal epi­ cells a reconstitution of the corneal epithelium is possible thelial phenotype.65 if enough early transient amplifying cells are present. The technique of conjunctival transplantation has When transient amplifying cells are lost simultaneously earlier been suggested for the surgical reconstruction of with the surgical depletion of corneal stem cells 96% of vascularised corneas after chemical burns in human the eyes show corneal neovascularisation and expression patients.66-6s To compare this method which was based on of a conjunctival phenotype in the epithelium which the hypothesis that corneal epithelium derives from con­ covers the cornea.31 This result confirms that absence of junctival epithelium with limbal transplantation Tsai and both stem and transient amplifying cells does not allow the coworkers65 used both methods for the treatment of reconstitution of a genuine corneal epithelium due to the experimental Iimbal deficiency. The results showed that total loss of the proliferative reserve. In the absence of the all but one of eight corneas which were treated by con­ Iimbal proliferative barrier conjunctival epithelium grows junctival transplantation developed progressive vascular­ onto the cornea and retains its phenotype in the presence isation with continuous expression of a conjunctival of fast-developing neovascularisation. phenotype. A comparison with the above results after lim­ Taken together the results of these animal experiments bal transplanation confirms the importance of corneal stress the importance of both corneal stem cells and tran­ stem cells at the limbus for the surgical reconstitution of sient amplifying cells for the maintenance and regener­ the corneal epithelial phenotype. ation of the corneal epithelium. While both popUlations in the Iimbal epithelium easily maintain and regenerate the REGULATION OF CORNEAL STEM AND central and peripheral corneal epithelium, a partial or total TRANSIENT AMPLIFYING CELLS loss of stem cells at the limbus can only be compensated if Most of the current knowledge concerning epithelial stem a sufficient amount of early transient amplifying cells cells is derived from investigations of haematopoietic pro­ remains in the corneal epithelium. In quantitative terms, it genitor cells, which were made possible by the develop­ seems that up to 50(/c of the Iimbal epithelium is lost with­ ment of single-cell clonal growth assays such as the spleen out conjunctival ingrowth and vascularisation if a suf­ colony forming assay or the agarose assay.69,70 These ficient amount of transient amplifying cells remains. model systems allow the development of single progenitor These experimental findings were also confirmed in cells to be traced and their proliferative behaviour at dif­ human patients, where maintenance and healing of cor­ ferent stages of the proliferative cascade to be neal epithelium is also dependent on the integrity of the investigated. limbus. Clinical studies of patients with chemical burns In order to gain insight into the regulation of the pro­ observed rapid, uncomplicated epithelial wound healing liferation of corneal progenitor cells we developed a when the inj ury concerned only the central and peripheral similar single-cell clonal growth assay which allows corneal epitheli um.IO When the corneal damage extended single corneal epithelial cells to be cultured in a serum­ to the limbus, small defects in the limbal circumference free defined culture medium.os.?1 As mentioned before, are closed by sliding of the adjacent limbal epithelium, a presently available methods for the assessment of differ­ process which reconstitutes the limbal barrier. If larger entiation or proliferation do not permit differentiation portions of the limbal circumference are destroyed (1800 between corneal stem and transient amplifying cells in a or more), the lack of stem and transient amplifying cells culture dish. However, the unique differential location of results in ingrowth of conjunctival epithelium as well as in these cells in the limbal and corneal epithelium allows the regional neovascularisation. 10 Animal studies suggest that separate investigation of these progenitor cell subpopula­ this kind of defect of the Iimbal barrier against the con­ tions. Anatomical separation of the limbal and corneal junctival epithelium also occurs when the entire corneal epithelium results in one cell population which contains epithelium is missing and the suprabasal limbal epithe­ both stem and transient amplifying cells (from the limbus) lium is removed.63 and a second cell population which contains only transient amplifying cells (from the cornea). A comparison of these Transplantation of Corneal Stem Cells populations in serum-free media showed a significant On the basis of the pathophysiological concept that a sim­ difference in the formation of colonies and their rate of ultaneous loss of corneal stem and transient amplifying proliferation, these being lower in limbal than in corneal cell causes alterations of the corneal phenotype, Kenyon cultures.sx This result, which is supported by labelling and Tseng64 suggested that the original corneal phenotype experiments in rabbits,S can be explained by the slow could be reconstituted by transplanation of healthy cor­ cycling nature of stem cells in the limbal cell popUlation neal stem cells. The effectiveness of this method for the and the lower number of fastcycling transient amplifying treatment of experimentally induced limbal deficiency in cells in limbal cultures. Furthermore, investigations of the rabbits was investigated by Tsai and coworkers.60 In all differentiation ofthe colonies suggest that the culture con­ eight eyes in which a simultaneous removal of limbal and ditions predominantly promote the proliferation of tran­ corneal epithelium had caused aplasia of limbal stem cells sient amplifying cells.58 with conjunctivalisation and neovascularisation, limbal A survey of various polypeptide growth factors such as 176 F. E. KRUSE

epidermal growth factor (EGF) basic and acidic fibroblast stem and transient amplifying cells (Fig. 2). The differ­ growth factor (a and bFGF) or nerve growth factor (NGF) entiation of stem to transient amplifying cells seems to be showed that these mitogens stimulate the proliferation of stimulated by serum factors one of which could be retinoic limbal and corneal epithelial cells in a similar way, indi­ acid. The amplification of transient amplifying cells cating that they promote the common subpopulation of seems to be supported by polypeptide growth factors such transient amplifying cells.72 The proliferation of transient as EGF, aFGF, bFGF or NGF and increasing concen­ amplifying cells was also promoted by increasing concen­ trations of extracellular calcium. The amplification of trations of extracellular calcium.73 In contrast to these transient amplifying cells seems to be inhibited by TGF-� stimulatory effects, transforming growth factor-beta and serum factors such as retinoic acid. (TGF-�) inhibited the proliferation of transient ampli­ The physiological meaning of some of these regulatory fying cells in both limbal and corneal cultures.72 factors can be illustrated by the differential nutritional While polypeptide growth factors and calcium also supply of limbal and corneal epithelium. The limbal epi­ caused an almost identical pattern of differentiation in thelium is under the influence of the Iimbal vasculature both limbal and corneal cultures, which supports their while the corneais avascular.8 1.82 Therefore, factors from primary effect on transient amplifying cells, the addition serum such as retinoic acid might occur in limbal epi­ of fetal bovine serum (FBS) resulted in a different thelium in higher concentrations than in corneal epi­ response. While colony formation and proliferation of thelium and the resulting concentration gradients could corneal cultures was reduced by increasing concentrations have regulatory functions.Furthermore, most of the poly­ of FBS, the formation of limbal colonies and their pro­ peptide growth factors investigated by us have been liferation was stimulated.74 These results indicate that fac­ proved to exist in the limbal and corneal epithelium.83-88 tors in serum allow the preferential stimulation of a Interestingly, a recent investigation by Li and Tseng89 subpopulation of limbal epithelial cells which might described significant regional differences concerning the represent stem cells while on the other hand transient distribution of growth factors in human limbal and corneal amplifying cells in the corneal epithelium were inhibited. epithelium as well as underlying stroma. Several of the This interpretation was further supported by the obser­ vation of several morphologically distinct subsets of colo­ EGF, nies present in limbal and corneal cultures. Due to the rapid differentiation of cells under the culture conditions aFGF, bFGF employed in these studies the colonies most probably con­ tained different types of transient amplifying cells some of Serum, NGF, which were the product of a recent conversion from stem cells.74 Retinoic Calcium One of the factors in serum which might be responsible acid for the presumed differentiation of limbal stem cells in ret­ inoic acid. Retinoic acid is an important modulator of epi­ thelial proliferation and differentiation which is present in + serum in biologically active concentrations.75.76 Further­ more, the results of experiments with embryonal stem cells suggest that retinoic acid induces the differentiation of stem cells.77-79 Both limbal and corneal cultures GC-TAUPMC exhibited a biphasic patternof proliferation in response to increasing concentrations of retinoic acid.80 While high concentrations above 1O-6M inhibited the formation of colonies in both limbal and peripheral corneal cultures, lower concentrations had a differential effect on these cell populations. The colony formation in limbal cultures was Serum, increased by retinoic acid (10-8 or 1O-7M) but the colony formation in corneal cultures remained unchanged.so This Retinoic differential stimulation of colony formation in limbal cul­ acid, tures allows us to speculate that limbal stem cells were preferentially differentiated to transient amplifying cells TGF-B 1 under the influenceof retinoic acid. Furthermore, cultures containing 10-8M retinoic acid allowed the identification of morphologically distinct colony types, one of which Fig. 2. Model of the regulation of corneal stem and transient was almost exclusively observed in limbal cultures.80 This amplifying cells. Conversion of stem cells (SC) to transient result further supports the notion that retinoic acid acts on amplifying cells (TAC) is supported by serum factors such as ret­ inoic acid. Amplification of TAC is promoted by epidermal a special subpopulation of progenitor cells in the limbal growth factor (EGF), acidic and basic fibroblast growth factor epithelium, i.e. stem cells. (a and b FGF) and nerve growth factor (NGF) and calcium. Taken together our investigations permit the construc­ Amplification ofTAC is inhibited by retinoic acid and transfor­ tion of a hypothetical model of the regulation of corneal ming growth factor-beta (TGF-�l) PMC, post-mitotic cells. STEM CELLS OF THE CORNEAL EPITHELIUM 177 investigated factors (e.g. members of the EGF and FGF CLINICAL APPLICATION family of growth factors) were expressed in the corneal Although experimental evidence of the existence of cor­ epithelium while their receptors were observed in the . neal epithelial stem cells in the limbus is derived exclus­ und er I ymg stroma or vice. versa.89 Taken together, physio- ively from animal studies it seems likely that human logically occurring regional differences in various regu­ corneal stem cells are also located only in the basal limbal latory factors might play a role in the regulation of limbal epithelium. This assumption is based on three lines of evi­ and corneal epithelium. dence: First, investigations of cellular differentiation Besides regional concentration gradients of various fac­ show that the human basal limbal epithelium also lacks the tors, 'cross-talk' between different cell populations such expression of differentiation-related keratins (such as as corneal epithelial and stromal cells might also have keratin K3) and expresses markers indicative of an regulatory functions. The potential importance of mes­ undifferentiated phenotype (such as keratin K19, vimentin enchymal cells such as embryonal fibroblasts for the regu­ or alpha-enolase).48-so,lol Second, the conjunctival pheno­ lation of epithelial cells can be concluded from a number type which is expressed by corneas afterthe simultaneous of experiments. Using inactivated embryonal mouse loss of the corneal and limbal epithelium can be reconsti­ fibroblasts (3T3 feeder layer) Rheinwald and Green90 tuted into a corneal phenotype by transplantation oflimbal were able to culture single keratinocytes for multiple pass­ stem cells.64,102 Third, circumstantial evidence for the age. Application of this technique for clinical use makes it existence of corneal stem cells is provided by the obser­ possible to expand the skin of burn victims and to regraftit vation that neoplasms of the corneal epithelium, which successfully onto the patient.91,91 The long-term survival can be interpreted as malfunctions of stem cells, almost of the graftsin recipients suggests that these culture condi­ always originate from the limbal epithelium. 103,lo4 tions preserve epithelial stem cells. This interpretation is The concept of the location of human corneal epithelial further supported by experiments which show that 3T3 stem cells in basal limbal epithelium can be used to cells also allow the culture of haematopoietic stem cells.93 improve the pathophysiological interpretation of various We and others were able to demonstrate that 3T3 cells also allow the culture of corneal epithelial cells.94-96 A com­ disorders of the ocular surface epithelium as well as to parison of limbal cultures suggested that 3T3 cells pro­ design therapeutic concepts for such diseases. mote the proliferation of a subpopulation of cells which is Ocular Surface Disorders Caused by Malfunction present only in the limbal epithelium.9s,96 Furthertnore, or Absence of Corneal Stem Cells 3T3 cells selectively allow the long-term culture of human The absence or malfunction of corneal stem cells is limbal epithelial cells but not of corneal epithelial cells.97 characterised by the loss of the proliferative capacity of These results suggest that factors from 3T3 cells may pre­ the corneal epithelium. Presumably due to the breakdown serve stem cells or early transient amplifying cells. of the proliferative barrier at the limbus, conjunctival epi­ One of the mechanisms by which 3T3 cells might thelium can grow onto the corneal surface and may be sup­ modulate epithelial cells could be a change of their ported by newly fortned vessels. The resulting clinical response to factors such as TGF-� which inhibit epithelial picture, which was initially described in humans by cell proliferation and promote differentiation.98 As men­ Tseng,40 can be characterised by analogy to the experi­ tioned above, TGF-� inhibited the colony formation and mental removal of corneal stem cells in animal models. proliferation and induced differentiation in limbal and The clinical symptoms may include decreased vision, corneal epithelial cultures.99 However, the addition of 3T3 cells completely changes this response and actually pro­ photophobia, tearing, blepharospasm and recurrent epi­ moted colony formation and proliferation in response to sodes of pain, as well as a history of chronic inflammation TGF-� in concentrations up to l.0 ng/m1.99 Although this with redness and oedema. The biomicroscopical findings modulation needs further exploration a possible mech­ at slit lamp examination may include a dull and irregular anism is the secretion of stimulatory factors by 3T3 cells. reflex of the corneal epithelium which is variable in thick­ Since such hypothetical factors should counteract differ­ ness and opacification. The deep layers of the epithelium entiation and pertnitthe amplification of corneal epithelial and anterior stroma may contain blood vessels and areas of progenitor cells, their isolation could be of great thera­ opacification. Severe malfunction or absence of corneal peutic importance. As a first step towards the identi­ epithelial stem cells may result in an ingrowth of thick­ fication of such factors, we have recently shown that the ened fibrovascular pannus as well as in calcifications. His­ conditioned medium from 3T3 cells contains factors tologically the ingrowth of conjunctival epithelium has which promote colony formation and proliferation of lim­ been documented by the presence of goblet cells in ! bal and corneal epithelial cultures and that such factors impression cytology. Os Immunohistochemically both the possibly could be isolated from conditioned medium.lOo absence of a cornea-type differentiation (such as the Taken together these results suggest that stromal fibro­ absence of keratin K3) as well as the presence of mucin in blasts play a physiological role in the regulation of limbal goblet cells has been shown by monoclonal , and corneal epithelial cells and highlight the interplay antibodies. 102 106 between various regulatory factors as well as cell types in The aetiology of the absence or malfunction of stem the regulation of corneal stem and transient amplifying cells can be classified as primary or secondary (Table I). cells. Primary insufficiency of the limbal epithelium can be 178 F. E. KRUSE

Table I. Aetiology of Iimbal insufficiency epithelium can also cause permanent dysfunction of lim­ bal stem cells. Examples are excision of limbus-based sus­ Primary; Anirida Congenital erythrokeratodermia picious neoplastic malformations such as squamous cell

Secondary; Chemical injury carcinoma or corneal intra-epithelial neoplasms, excision Thermal injury of recurrent pterygium and cryosurgery of the ciliary body Contact lens wear which is performed in the proximity of the limbus. The Limbal surgery resulting dysfunction of the limbal epithelium has suc­ cessfully been treated by autologous limbal trans­ characterised by the absence of external factors such as plantation.64.121 injuries, mechanical damage or pharmaceutical drugs. Although rare, two disease entities can be attributed to a Treatment (d"Depletion or Malfunction of Corneal loss or functional impairment of stem cells one of which is Stem Cells b,v Limbed Transplantation aniridia and the other congenital erythrokeratodermia. As long as limited insight into the regulation of prolifer­ Many patients with aniridia express irregular and cloudy ation and differentiation of limbal stem cells prevents the epithelium with corneal neovascularisations. By means of impression cytology goblet cells have been observed in development of a conservative treatment. surgical inter­ the superficial corneal epithelium, which suggests an vention serves as the only means of reconstituting the ingrowth of conjunctival epithelium due to loss of the lim­ ocular surface following disorders caused by limbal insuf­ bal barrier.los.l07 In congenital erythrokeratodermia, which ficiency. However, conservative surgical treatment of lim­ was initially described by Burns,108 a clear but irregular bal insufficiency, for example. after chemical burns. by procedures such as lamellar or perforating keratoplasty corneal epithelium is traversed by blood vessels of seem­ 22 2 ingly conjunctival origin. We have recently observed gob­ generally has a very poor prognosis.1 1 6 This can be let cells in the corneal epithelium of one patient with this explained by the fact that transplantation of the central or rare disorder, which also suggests a loss of the limbal pro­ peripheral corneal epithelium results only in the sub­ stitution of transient amplifying cells with a limited pro­ liferative barrier possibly due to an impairment of corneal 2 stem cells.109 liferati ve capacity and life span.1 6 Therefore, The majority of ocular surface disorders which are conventional keratoplasty provides only a temporary 2 caused by the absence or dysfunction of corneal stem cells replacement of the host's corneal epitheliumX.l 7 and does are of secondary origin. Most importantly, chemical and not permanently reconstitute the limbal barrier. In order to thermal burns can cause variable limbal epithelial damage improve the limited success of the conservative surgical and ischaemia of the limbal vasculature. The extent of this treatment of chemical burns Thoft66 68 initially suggested damage is the cornerstone of various classifications the procedure of conjunctival transplantation. This ll- regarding the prognosis after acute injury. I I I 13 Increased method was developed on the theory of a conjunctival permeability of the limbal vasculature as well as cellular derivation of the corneal epithelium. On the basis of the damage lead to an influxof leucocytes into the epithelium stem cell model Kenyon and TsengM further modifiedthe and stroma which alters the regulation of cellular prolifer­ technique of conjunctival transplantation to include the ation and differentiation.III.114-116 In contrast to minor limbal epithelium . Thus, the surgical procedure of limbal injuries in which a loss of corneal epithelium is combined transplantation results in transfer of limbal epithelium with minor limbal damage, larger defects of the limbal cir­ containing long-living corneal stem cells. cumference cannot heal by sliding of the adj acent healthy The principle of the procedure, which is shown in limbal epithelium.lo In these cases, and after injuries in Fig. 3, involves first the removal of altered corneal epi­ which severe inflammation resulted in regulatory dys­ thelium with removal of the pannus down to the bare function of the limbal epithelium, a localised loss of the sclera. Then a superficial keratectomy can be made in the limbal barrier occurs with consecutive invasion of con­ periphery of the cornea extending into the sclera to create , . junctival epithelium. 10 15 11 5,11 7 a bed for the limbal graft. The graft is then harvested from A combination of mechanical, toxic and inflammatory the donor eye in the form of two stripes of tissue each factors with hypoxia may lead to limbal insufficiency in measuring about 4 clock-hours (for autografts ). In the case contact lens wearers.118.119 The clinical triad of corneal of an allograft transplantation a ring graft containing 3600 neovascularisation, epithelial abnormalities (such as indo­ of limbal tissue can be prepared from the donor eye as lent ulceration and irregularities with whorled pattern) and shown in the illustration. In the light of the importance of stromal opacities has been called contact-lens-related epi­ the limbal and corneal stroma for the regulation of the cor­ thelial dysfunction.120 Although there are no published neal epithelium a sufficient portion of limbal stroma histological investigations regarding eyes with this dis­ should be included in the graft. The graft is then transfered order, which is unresponsive to all conservative treatment, onto the host and fixed at its corneal, scleral and con­ the successfulre constitution of the ocular surface by auto­ junctival margins. From this ring of limbal tissue transient logous limbal transplantation64.120 further supports the idea amplifying cells are generated which migrate onto the that a dysfunction of limbal stem cells is the key patho­ denuded corneal surface of the host. After successful genetic factor in the development of this disorder. transplantation the host's cornea wi II be permanently Several surgical manipUlations in the area of the limbal covered by epithelium from the donor. STEM CELLS OF THE CORNEAL EPITHELIUM 179

Pannus removal Limbal graft Under this therapeutic regimen long-term rehabilitation � of bilateral limbal insufficiency can be achieved.'33 The eye of a representative patient with partial limbal insuf­ ficiency due to multiple surgery for recurrent pterygium is shown prior to Iimbal allograft transplantation with epi­ thelial opacities and superficial as well as deep stromal vessels in Fig. 4 (above). One year after limbal allograft e-/�- transplantation in both nasal quadrants the limbal epi­ , thelium is restored by a zone of clear cornea without recur­ I .... , , rence of the vessel ingrowth (Fig. 4, below). The central I \ I \ I I epithelium, which was removed to the level of Bowman's I I \ , layer, was rapidly substituted from the grafted epithelium. , I ....'0,.,.-----, .... , '.... ' Furthermore, the opacifiedstroma which was not touched .... _- ---", by the surgery cleared remarkably over the course of the Fig. 3. Principle of limbal transplantatioll. The procedure year, resulting in an improvement in visual acuity from involves the fo llowing steps; removal of fibrovascular tissue 1/50 to 20/35. Such a remarkable diminution of stromal (r om host, preparation (�(limbal graf t f/"om dOllar, tramie r ()j' opacities was also observed after limbal autograft trans­ graft to host and fixation at limbus. Reconstitution of host 's cor­ neal phenotype from donor occurs fo llowing successjit! plantation64 and might be due to an epithelial-stromal transplantatioll. interaction. However, in most cases a second surgical intervention such as lamellar or penetrating ketratoplasty In their original report Kenyon and TsengM described is needed for complete visual rehabilitation. In summary, 26 Iimbal autografts, 20 of which resulted in stable epi­ the procedure of limbal allograft transplantation as thelium without recurrent erosions and 15 in regression of initially described by Tseng and coworkers1 30 has proved the corneal neovascularisation. Since their original report to be a valuable tool in the reconstruction of bilateral lim­ the procedure of Iimbal autograft transplantation has bal insufficiencies. gained acceptance and several small series have been reported with good reconstitution of the corneal epithe­ CONCLUSION lium and regression of neovascularisation.'2o.'2x Although A wide body of experimental and clinical data support the autologous transplantation has significantly enhanced the model of the limbal location of corneal epithelial stem surgical armamentarium for treatment of unilateral limbal cells. Although the importance of these long-lived pro­ insufficiency, it is not applicable in bilateral cases which genitor cells for the integrity and regeneration of the cor­ frequently develop after injuries such as chemical burns. neal epithelial phenotype is evident, several open Such patients could be treated by heterologous limbal questions remain. One of these concerns the positive iden­ transplantation. However, this procedure carries a signifi­ tification of stem cells. Although several antibodies have cant risk of graft rejection because the Iimbal tissue is been developed neither stem cells, transient amplifying grafted into a highly vascular bed which does not enjoy the cells nor post-mitotic cells can currently be differentiated. immunological privilege of the avascular cornea. Since Also, almost nothing is known about the regulation of the Weise and coworkers'29 demonstrated the feasibility of self-renewal of stem cells and their resistance to differ­ conjunctival allografts in humans several years ago, it entiation-inducing agents. The lack of suitable model seemed justifiable also to perform Iimbal allografts in systems as well as the enormous complexity of the regu­ patients with bilateral legal blindness due to severe bilat­ latory mechanisms have until now precluded investigation eral limbal deficiency. Although no clinical study has been of these important questions. However, an increasing published, initial experience by Tseng disclosed a 30% number of cytokines are currently being studied and inves­ 130 chance of graft rejection in non-tissue-matched grafts. tigation of the interaction of various cell types (such as Subsequent use of oral cylosporin A in addition to oral epithelial and stromal cells) might eventually enable the steroids by Tsai and coworkers has significantly reduced identification of an environmental niche which governs the risk of allograft rejections and allowed successful the regulation of stem cells. The identification of factors graftsin 12 patients.IJO Similar to the use of oral cyclospo­ which prevent the differentiation of stem cells and allows rin A in perforating keratoplasty, where the drug is admin­ their amplification has enormous clinical potential. Such istered for a minimum of 6- 12 months,U' patients with a factors would allow for a conservative treatment of ocular limbal transplant should be treated for an equally long surface disorders which are due to stem cell loss or dys­ period. Since the long-term use of cyclosporin A can function. Therefore, these factors would overcome the result in serious side effects we evaluated the local admin­ limitations of the current treatment of ocular surface dis­ istration of the drug concerning the concentration in Iim­ orders with growth factors. These limitations can be bal tissue. First results of this study showed no significant extrapolated from the regulatory model shown in Fig. 2. difference in the concentration after systemic or topical All the factors which are included in this model promote administration."2 It therefore seems justfiable to di scon­ the differentiation of either stem cells or transient ampli­ tinue oral cyclosporin after a period of 12 weeks and to fying cells and therefore cannot be used for the treatment continue topical medication for another 9 months. of limbal insufficiency. 180 F. E. KRUSE

Fig. 4. Case report of limbal allograft transplantation. Limbal insufficiency of both nasal quadrants due to multiple surgery prior to current surgery(l eft upper quadrant). Central corneawith opacijicationand neovascularisation (right upper quadrant). Nasally located graft J year after limbal allograft (left lower quadrant). Central cornea after surgery with resolution of the deep opacijications in the avascular cornea (right lower quadrant).

The investigation of corneal sterncells has significantly REFERENCES enhanced our understanding of ocular surface disorders l. Buschke W, Friedenwald JS, Fleischmann W. Studies on and has led to the development of new therapeutic pro­ the mitotic activity of the cornealepithelium. Johns Hop­ cedures. In particular bilateral limbal insufficiencies, kins Hosp Bull 1943;72;143. 2. Kaufmann B, Gay H, Hollaender A. Distribution of mostly due to chemical burns, can be successfully rehabil­ mitoses in the cornealepithelium of the rabbit and the rat. itated by limbal allografts. However, more experience Anat Rec 1944;90;16l. with this procedure needs to be gathered, especially con­ 3. Hanna C, O'Brien JE. Cell proliferation and migration in cerning the signs of graft rejection, which are presently ill the epithelial layer of the cornea.Arch Ophthalmol 1960; defined. It is to be hoped that the enhancement of our basic 64;536. 4. Hanna C, Bicknell DS, O'Brien JE. Cell turnover in the knowledge concerning stern cell perpetuation might adult human eye. Arch Ophthalmol 1961;65 ; Ill. eventually allow us to expand single stern cells in culture 5. Lavker RM, Dong G, Cheng SZ, Kudoh K, Cotsarelis G, and to use them for reconstitution of the ocular surface. Sun TT. Relative proliferative rates of limbal and corneal epithelia. Invest Ophthalmol Vis Sci 1991;32;1864. This investigation was supported in part by a grant from 6. Kuwabara T, Perkins DG, Cogan DG. Sliding of the epi­ Dentsche Forschungsgemeinschaft (Kr 99311-1). thelium in expelimental corneal wounds. Invest Ophthal­ mol Vis Sci 1976;15;4. Key words: Cornea. Epithelium. Growth factors, Keratoplasty, Limbus. 7. Buck RC. Cell migration and repair of mouse cornea epi­ Stem cells. thelium. Invest Ophthalmol Vis Sci 1979;18;767. STEM CELLS OF THE CORNEAL EPITHELIUM 181

8. Kinoshiota S, Friend J, ThoftRA. Sex chromatin of donor differentiation induced by systemic vitamin A deficiency corneal epithelium in rabbits. Invest Ophthalmol Vis Sci in vascularized rabbit corneas. Invest Ophthalmol Vis Sci 1981 ;21 ;434. 1987;28; 1497. 9. Buck RC. Measurement of centripetal migration of normal 32. Cintron C, Hasslinger L, Kublin CL, Friend J. A simple corneal epithelium in the mouse. Invest Ophthalmol Vis method for removal of rabbit corneal epithelium utilizing Sci 1985;26;1 296. n-heptanol. Ophthalmic Res 1979; 11;90. 10. Dua HS, Forrester JY.The corneoscleral limbus in human 33. Kruse FE, Chen JJY, Tsai RJF, Tseng SCG. Conjunctival corneal epithelial wound healing. Am J Ophthalmol 1990; transdifferentiation is due to incomplete removal of limbal 110;545. basal epithelium. Invest Ophthalmol Vis Sci 1990;31; 11. Kaye DB. Epithelial response in penetrating keratoplasty. 1903. Am J Ophthalmol 1980;89;381. 34. Davanger M, Evensen A. Role of the pericorneal peri­ 12. Lemp MA, Mathers WD. Cornealepithelial cell movement papillary structure in renewal of cornealepithelium. Nature in humans. Eye 1989;3;438. 1971 ;229;560. 13. Thoft RA, Friend J. The X, Y, Z hypothesis of cornealepi­ 35. Schermer A, Galvin S, Sun TT. Differentiation related thelial maintenance. Invest Ophthalmol Vis Sci 1983;24; expression of major 64K keratin in vivo and in culture sug­ 1442. gests limbal location of cornealepithelial stem cells. J Cell 14. Sharma A, Coles WHo Kinetics of corneal epithelial main­ Bioi 1986;103;49. tenance and graftloss. Invest Ophthalmol Vis Sci 1989;30; 36. Potten CS. Epithelial proliferative subpopulations. In; 1962. Lord BI, Potten CS, Cole RJ, editors. Stem cells and tissue 15. Mann I, Pullinger BD. A study of mustard gas lesions of the homeostasis. Cambridge; Cambridge University Press, eyes of rabbits and man. Proc R Soc Med 1942;35;229. 1978;317. 16. Maumanee AE, Scholz RO. Histopathology of the ocular 37. Lebold CP. The life history of cells in renewing systems. lesions produced by sulfur and nitrogen mustard. John Am J Anat 1981;160;114. Hopkins Hosp Bull 1948;82;121. 38. Lathja LG. Stem cell concepts. Differentiation 1979; 14;23. 17. Friedenwald JS, Buschke W, Scholz RO. Effects of mus­ 39. Lavker RM, Cotsarelis G, Dong G, Hardy C, Schermer A, tard and nitrogen mustard on mitotic and wound healing Sun TT. Limbal location of corneal epithelialst em cells. In; activities of the corneal epithelium. Johns Hopkins Hosp HD Cavanagh, editor. The cornea. Transactions of the Bull 1948;82;148. world congress on the cornea III. New York; Raven Press, 18. Friedenwald J. Growth pressure and metaplasia of con­ 1988;23. junctival and corneal epithelium. Doc Ophthalmol 1951; 40. Tseng SCG. Concept and application of limbal stem cells. 5-6;184. Eye 1989;3; 141. 19. Friend J, Thoft RA. Functional competence of regeneration 41. Spangrude GJ, Heimfeld S, Weissman IL. Purificationand ocular surface epithelium. Invest Ophthalmol Vis Sci 1978; characterization of mouse hematopoietic stem cells. 17;134. Science 1989;24 1 ;58. 20. Shapiro MS, Friend J, Thoft RA. Corneal re-epithelial­ 42. Moll R, Franke WW, Schiller DL, Geider B, Krepler R. ization from the conjunctiva. Invest Ophthalmol Vis Sci The catalog of human cytokeratins; patterns of expression 1981;21 ;135. in normal epithelia, tumors and cultured cells. Cell 1982; 21. Kinoshita S, Kiorpes TC, Friend J, Thoft RA. Limbal epi­ 31; 11. thelium in ocular surface wound healing. Invest Ophthal­ 43. Cooper D, Schermer A, Sun TT. Classification of human mol Vis Sci 1982;23;73. epithelia and their neoplasms using monoclonal antibodies 22. Kinoshita S, Friend J, Thoft RA. Biphasic cell proliferation to keratins; strategies, applications, and limitations. Lab in transdifferentiation of conjunctival to corneal epithe­ Invest 1985;52;243. lium in rabbits. Invest Ophthalmol Vis Sci 1983 ;24; 1008. 44. Tseng SCG, Jarvinen MJ, Nelson WG, Hunag J, Wood­ 23. Buck RC. Ultrastructure of conjunctival epithelium replac­ cock-Mitchell, Sun TT. Correlation of specific keratins ing cornealepit helium. Curr Eye Res 1986;5;149. with different types of epithelial differentiation; mono­ 24. ThoftRA, Friend J. Biochemical transformation of regen­ clonal antibody studies. Cell 1982;30;361. erating ocular surface epithelium. Invest Ophthalmol Vis 45. Woodcock-Mitchell JR, Eichner WG, Nelson WG, Sun Sci 1977;16;14. TT. Immunolocalization of keratin polypeptides in human 25. Kinoshita S, Friend J, Kiorpes TC, Thoft RA. Keratin-like epidermis using monoclonal antibodies. J Cell Bioi 1988; proteins in cornealand conjunctival epithelium are differ­ 95;580. ent. Invest Ophthalmol Vis Sci 1983;24;577. 46. Chaloin-Dufau C, Sun TT, Dhouailly D. Appearance of the 26. Harris TM, Berry ER, Pakurar AS, Sheppard LB. Bio­ keratin pair K3/K 12 during embryonic and adult corneal chemical transformation of bulbar conjunctiva into corneal epithelial differentiation in the chick and in the rabbit. Cell epithelium; an electrophoretic analysis. Exp Eye Res 1985; Differ Dev 1990;32;97. 41;597. 47. Kurpakus MA, Stock EL, Jones JCR. Expression of the 55- 27. Huang AJW, Watson BD, Hernandez E, Tseng SCG. kD/64-kD corneal keratins inocular surface epithelium. Induction of conjunctival transdifferentiation on vascular­ Invest Ophthalmol Vis Sci 1990;31 ;448. ized corneasby photothrombotic occlusion of cornealneo­ 48. Kasper M, Moll R, Stosiek P, Karsten U. Patternsof cyto­ vascularization. Ophthalmology 1988;95;228. keratin and vimantin expression in the human eye. Histo­ 28. Tseng SCG, Hatchell D, TierneyN. Expression of specific chemistry 1988;89;369. keratin markers by rabbit corneal, conjunctival and eso­ 49. Lauweryns B, van den Oord JJ, De Vos R, Missotten L. A phageal during vitamin A deficiency. J Cell Bioi 1984;99; new epithelial cell type in the human cornea. Invest Oph­ 2279. thalmol Vis Sci 1993;34; 1983. 29. Farazdaghi M, Lui SH, Rider AA, Tseng SCG. Reversal of 50. Zieske JD, Bukusoglu G, Yankauchas MA. Character­ conjunctival transdifferentiation by retinoids. Invest Oph­ ization of a potential marker of corneal epithelial stem thalmol Vis Sci 1984;25;76. cells. Invest Ophthalmol Vis Sci 1992;33; 143. 30. Tseng SCG, Hirst LW, Faradzaghi M, Green WR. Inhibi­ 51. Zieske JD, Bukusoglu G, Yankauchas MA, Keutmann HT. tion of conjunctival transdifferentiation by topical ret­ A 50-kD protein present in limbal basal cells of limbal epi­ inoids. Invest Ophthalmol Vis Sci 1987;28;538. thelium is enolase. Invest Ophthalmol Vis Sci 1991;32; 31. Tseng SCG, Faradaghi M, Ryder AA. Conjunctival trans- 1073. 182 F. E. KRUSE

52. Withers RH. Colony fonning units in the intestine. In; clonal growth and differentiation of cultured limbal and Cairnie AB, Lala PK, Osmond DG, editors. Stem cells of corneal epithelium. Invest Ophthalmol Vis Sci 1993;34; renewing cell populations. New York; Academic Press, 2976. 1976;33. 75. Fuchs E, Green H. Regulation of terminal differentiation of 53. Bickenbach JR. Identification and behaviour of label­ cultured human keratinocytes by vitamin A. Cell 1981;25; retaining cells in oral mucosa and skin. J Dent Res 1981; 617. 60;1611. 76. Hashimoto T, Dykes PJ, Marks R. Retinoic acid-induced 54. Cotsarelis G, Cheng S-Z, Dong G, Sun TT Lawker RW. inhibition of growth and reduction of spreading of human Existence of slow-cycling limbal basal cells that can be epidermal cells in culture. Br J Dermatol 1985;112;637. preferentially stimulated to proliferate; implications on 77. Strickland S, Mahavi V. The induction of differentiation in limbal stem cells. Cell 1989;57;201. teratocarcinoma stem cells by retinoic acid. Cell 1978; 15; 55. Yuspa SH, Ben TB, Hennings H, Lichti U. Divergent 393. response in epidermal basal cells exposed to the tumor pro­ 78. Breitman TR, Selonick S, Collins S. Induction of differ­ moter 12-0-tetradecanylphorbol-13-acetate. Cancer Res entiation of the human promyelocytic cell line (HL-60) by 1982;42;2344. retinoic acid. Proc Natl Acad Sci USA 1980;77;2936. 56. Hawley-Neslon P, Stanley JR, Schmidt J. Gullino M, 79. Sherman ML, Mattheai KI, Schindler J. Studies on the Yuspa SH. The tumor promoter, 12-0-tetradecanylphor­ mechanism of induction of embryonal carcinoma cell dif­ bol-13-acetate, accelerates keratinocyte differentiation and ferentiation by retinoic acid. Ann NY Acad Sci 198 1 ;359; stimulates growth of an unidentified cell type in cultured 192. human epidermis. Exp Cell Res 1982;137;155. 80. Kruse FE, Tseng SCG. Retinoic acid modulates clonal 57. Hesterberg TW, Manesse SC, Iglehart JD, Sanchez JH, growth and differentiation of cultured limbal and periph­ Boreiko CJ. Subpopulations of human bronchial epithelial eral corneal epithelium. Invest Ophthalmol Vis Sci 1993 (in press). cells in culture respond heterogeneously to 12-0-tetrade­ 81. Buskirk EM van. The anatomy of the limbus. Eye 1989;3; canylphorbol-13-acetate (TPA) and other modulators of 101. differentiation. Carcinogenesis 1987;8; 1511. 82. McCulley JP. The circulation of fluidat the limbus (flow 58. Kruse FE, Tseng SCG. A serum-free clonal growth assay and diffusion at the limbus). Eye 1989;3;114. for limbal, peripheral and central corneal epithelium. 83. Caruelle D, Groux-Muscatelli B, Gaudric A. Sestier C, Invest Ophthalmol Vis Sci 1991 ;32;2086. Casocas G, Caruelle JP, Barritault D. Immunological study 59. Kruse FE, Tseng SCG. A tumor promoter-resistant sub­ of acidic fibroblastgrowth factor (aFGF) distribution in the population is larger in limbal epithelial than in corneal epi­ eye. J Cell Biochem 1989;39; 117. thelium. Invest Ophthalmol Vis Sci 1993;34;2501. 84. Baudouin C, Fredj-Reygrobellet D, Caruelle JP, Barritault 60. Tseng SCG, Zhang SH. Slow cycling nature of limbal epi­ D, Gastaud D, Lapalus P. Acidic fibroblast growth factor thelium demonstrated by 5-FU resistance. Invest Ophthal­ distribution in nonnal eye and possible implication in mol Vis Sci 1992;33;1 176. ocular pathogenesis. Ophthalmol Res 1990;22;73. 61. Chen JJY, Tseng SCG. Corneal epithelial wound healing in 85. Adamis AP, Meklir B, Joyce NC. Immunohistochemical partial limbal deficiency. Invest Ophthalmol Vis Sci 1990; distribution of basic-fibroblastgrowth factor in the cornea. 31;13 01. Invest Ophthalmol Vis Sci 1991 ;32;955. 62. Huang AJW, Tseng SCG. Corneal epithelial wound healing 86. Wilson SE, He YG, Lloyd SA. EGF, basic FGF and TGF-� in the absence oflimbal epithelium. Invest Ophthalmol Vis messenger RNA production in rabbit corneal epithelial Sci 1991 ;32;96. cells. Invest Ophthalmol Vis Sci 1992;33; 1987. 63. Chen JJY, Tseng SCG. Abnormal cornealep ithelial wound 87. Wilson SE, He YG, Lloyd SA. EGF, EFG-receptor, basic healing in partial-thickness removal of Iimbal basal epi­ FGF, TGF-�l, and IL-Ia mRNA in human corneal epi­ thelium. Invest Ophthalmo1 Vis Sci 1991;32;22 19. thelial cells and stromal fibroblasts. Invest Ophthalmol Vis 64. Kenyon KR, Tseng SCG. Limbal autograft transplantation Sci 1992;33; 1756. for ocular surface disorders. Ophthalmology 1989;96;709. 88. Schultz GS, Khaw PT. MacKay SLD, Chegini N, Rotatori 65. Tsai RJP. Sun TT, Tseng SCG. Comparison of Iimbal and DS, Adams JL, Shimizu RW. Human corneal epithelium; conjunctival autograft transplantation in corneal surface autocrine production of transforming growth factor-alpha reconstruction in rabbits. Ophthalmology 1990;97 ;446. mRNA and protein. Invest Ophthalmol Vis Sci 1992;33; 66. Thoft RA. Conjunctival transplantation. Arch Ophthalmol 1028. 1977;95;1425. 89. Li DQ, Tseng SCG. Cytokine network between human Iim­ 67. Thoft RA. Conjunctival transplantation as an alternative bal or corneal epithelium and fibroblasts. Invest Ophthal­ for keratoplasty. Ophthalmology 1979;86: 1084. mol Vis Sci 1993;34;1011. 68. Thoft RA. Indications for conjunctival transplantation. 90. Rheinwald JG, Green H. Serial cultivation of strains of Ophthalmology 1982;89;335. human epidermal keratinocyte; the formation of keratiniz­ 69. Becker AJ, McCulloch EA. Till JE. Cytological demon­ ing colonies from single cells. Cell 1975;6;331. stration of the clonal nature of spleen colonies derived from 91. Green H, Kehinde 0, Thomas J. Growth of cultured epider­ transplanted mouse marrow cells. Nature 1963; 197 ;453. mal cells into multiple epithelia suitable for grafting. Proc 70. Plutznik DH, Sachs L. The cloning of normal 'mast' cells Natl Acad Sci USA 1979;76;5665. in tissue culture. ] Cell Comp Physiol 1965;66;3 19. 92. O'Connor N, Mulliken lB, Banks-Schlegel S, Kehunde 0, 71. Kruse FE, Tseng SCG. Zur Untersuchung des Limbusepi­ Green H. Grafting of burns with cultured epithelium pre­ thels in vitro. Fortschr Ophthalmol 1991 ;88; 107. pared from autologous epidermal grafts.Lanc et 1981;I 0; 72. Kruse FE, Tseng SCG. Growth factors modulate clonal 75. growth and differentiation of cultured rabbit limhal and 93. Roberts RA Spooncer E. Parkinson EK, Lord Bl, Allen corneal epithelium. Invest Ophthalmol Vis Sci 1993;34; TD, Dexter TM. Metabolically inactive 3T3 cells can sub­ 1963. stitute for marrow stromal cells to promote the prolifer­ 73. Kruse FE, Tseng SCG. Proliferative and differentiative ation and development of muItipotent haematopietic stem response of corneal and limbal epithelium to extracellular cells. J Cell Physiol 1987;132;203. calcium in serum-free clonal cultures. J Cell Physiol 1992; 94. Sun TT, Green H. Cultured cells of cornea,con junctiva and 151;347. skin; absence of marked intrinsic divergence oftheir differ­ 74. Kruse FE, Tseng SCG. Serum differentially modulates the entiated states. Nature 1977;269;489. STEM CELLS OF THE CORNEAL EPITHELIUM 183

95. Kruse FE, Tseng SCG. Comparison of single cell clonal 1 14. Brown SI, Wassermann HE, Dunn MW. Alkali burns of the growth of corneal and limbal epithelium in serum free cornea. Arch Ophthalmol 1969;82;91. definedcul ture and 3T3 feeder layer. I Cell Bioi 1989; I 09; 1 15. Duke-Elder S. System of ophthalmology. Vol. XIV, part 2. 329a. St Louis; Mosby. 1972;1011. 96. Tseng SCG, Kruse FE. Comparison of single cell clonal 116. Matsua H. Smelser GK. Epithelium and stroma in alkali­ growth of limbal and corneal epithelium in serum free burned corneas. Arch Ophthalmol 1973;89;396. defined medium and 3T3 feeder layer. Invest Ophthalmol 117. Faulkner W J, Kenyon KR. Rowsey Jl, Hanninen L, Gipson Vi s Sci 1990;3 1;538. IK, Gayler 1. Chemical burnsof the human ocular surface; 97. Lindberg K. Chaves HY. Shams NBK. Kenyon KR. Rhein­ clinicopathological studies in 14 cases. Invest Ophthalmol wald IG. Serial propagation of human ocular surface epi­ Vis Sci 1981;20(Suppl);8. thelial cells and reconstruction of differentiated epithelia. 118 . Schecter DR, Emery JM. Soper lW. Corneal neovascular­ Invest Ophthalmol Vis Sci 1992;33; 1 176. ization in therapeutic soft lens wear. Contact Intraocul 98. Masui T, Wakefield LM, Lechner JF, LaVeck MA, Sporn Lens Med I 1975;1;141. MB, Harris CC. Type beta transforming growth factor is 119. Weinberg RI. Deep corneal vascularization caused by the primary differentiation-inducing serum factor for aphakic soft contact lens wear. Am J OphthalmoI 1977;83; normal human bronchial epithelial cells. Proc Natl Acad 121. Sci USA 1986:83;2438. 120. Clinch TE, Goins KM. Cobo LM. Treatment of contact 99. Kruse FE. Tseng SCG. Inhibitory effect of TGF-� on cor­ lens-related ocular surface disorders with autologous con­ neal epithelial proliferation can be reversed by fi broblasts junctival transplantation. Ophthalmology 1992;99;634. coculture. Invest Ophthalmol Vis Sci 1992;33: I 028. 121. Copeland RR Jr, Char DH. Limbal autograft reconstruction 100. Kruse FE. Tseng SCG. Epithelial growth is promoted by a after conjunctival squamous cell carcinoma. Am J Oph­ mitogen from fibroblast conditioned medium. Invest Oph­ thalmoI 1990; 110 ;4 12. thalmol Vis Sci 1993;34; 101 1. 122. Brown SI. Tragkis MP, Pearce DB. Corneal transplantation 101. Thoft RA, Wiley LA, Sunderraj N. The multiponetial cells of severe alkali burns. Trans Am Acad Ophthalmol Otolar­ of the limbus. Eye 1989;3; I 09. yngol 1972;76; 1266. 102. Kenyon KR, Bulusoglu G, Zieske 10. Clinical pathologic 123. Brown SI, BloomfleldSE . Pearce DB. Follow-up report on correlations of limbal autografttran splantation. Am I Oph­ transplantation of the alkali burned cornea. Am J Ophthal­ thalmol 1990;3 1:I. mol 1974;77;538. 124. Abel R lr. Binder PS. Pi lack FM. Kaufman The results 103. Waring GO Ill, Roth AM, Elkins MB. Clinical and path­ HE. of penetrating keratoplasty after chemical burns. Trans Am ologic description of 17 cases of corneal intraepithelial Acad Ophthalmol Otolaryngol 1975:79;584. neoplasia. Am I Ophthalmol 1984:97 :547. 125. Kramer SG. Late numerical grading of alkali burns to 104. Garner A. The pathology of tumors at the limbus. Eye determine keratoplasty prognosis. Trans Am Ophthalmol 1989;3;210. Soc 1983:81 ;97. 105. Tseng SCG. Application of corneal impression cytology to 126. Kruse FE. Koch 1M. Klein U, Rohrschneider K. Volcker study conjunctival transdifferentiation defect. In; Orsoni HE. Waubke TN. The importance of limbal stem cells for JG, editor. Ophthalmic cytology. Proceedings of the first the therapy of chemical burns. Ger J Ophthalmol 1993;2; international symposium on ophthalmic cytology. Parma: 289. Centro grafica editoriale, Universita degli studi di Parma, 127. Lemp MA. The surface of the corneal graft; ill\'il'o colour 1988;65. specular microscopic study in the human. Trans Am Oph­ 106. Huang AJW. Tseng SCG. Development of monoclonal thalmol Soc 1989;87;6 19. antibodies to rabbit ocular mucin. Invest Ophthalmol Vis 128. Mashima Y. Yamada M. Yamada H, Tsunoda K. Arimoto Sci 1987;28;1483. H. Limbal autograft transplantations for chronic ocular 107. Nishida K. Kinoshita, Ohashi Y Kuwayama Y Ocular sur­ surface failures. Jpn J Clin Ophthalmol 1993;47:607. face abnormalities in aniridia. Ipn J Clin Ophthalmol 1993: 129. We ise RA. Mannis MJ. Vastine OW. Fujikawa LS. Roth 47;625. AM. Conjunctival transplantation; autologous and homol­ J 108. Burns FS. A case of generalized congenital keratodermia. ogous grafts. Arch Ophthalmol 1985; 103; 1736. Cutan Dis 1915;33;255. 130. Tseng SCG. Chen .TJY. Huang AJW. Kruse FE. Tsai RJ . 109. Kruse FE, Rohrschneider K, Blum M. Volcker HE. Tilgen Classification of conjunctival surgeries for corneal dis­ M, Longere G. Anton-Lamprecht I. Ocular findings in pro­ eases based on stem cell concept. Ophthalmol Clin North gressive erythrokeratodermia. Ger J Ophthalmol 1993;2; Am 1990:3;595. 368. 131. Sundmacher R, Reinhard T. Heering P. Six years' experi­ 1 10. Hughes WR Jr. Alkali burn s of the eye. 1. Review of the lit­ ence with systemic cyclosporin A prophylaxis in high-risk erature and summary of present knowledge. Arch Ophthal ­ perforating keratoplasty patients; a retrospective study. Ger mol 1946;35;423. J Ophthalmol 1992:1 ;432. Ill. Hughes WR JR. Alkali burns of the eye. II. Clinical and 132. Pfall B. Kruse FE. Klein U. Zorn M. Rohrschneider K. pathologic course. Arch Ophthalmol 1946;36; 189. Volcker HE. Comparison between local and systemic 112. Roper-Hall MJ . Thermal and chemical burns. Trans Oph­ cyclosporin A concerning the concentration in anterior thalmol Soc UK 1965;85;63 1. chamber and conjunctiva. Ger J Ophthalmol 1993;2;272. 113. Thoft RA. Chemical and thermal injury. Int Ophthalmol 133. Kruse FE, Rohrschneider K, Volcker HE. Limbal allografts Clin 1979; 1 9;243. [case reportl. Ger I Ophthalmol 1993;2;376.