Epithelial stem cells in the skin: definition, markers, localization and functions

Transcription

1 Exp Dermatol 1999: 8: Copyright C Munksgaard 1999 Printed in Denmark All rights reserved Controversies in Experimental Dermatology Section Editor: Ralf Paus, Berlin ISSN Epithelial stem cells in the skin: definition, markers, localization and functions Cotsarelis G, Kaur P, Dhouailly D, Hengge U, Bickenbach J. Epithelial stem cells in the skin: definition, markers, localization and functions. Exp Dermatol 1999: 8: C Munksgaard, 1999 G. Cotsarelis, P. Kaur, D. Dhouailly, U. Hengge and J. Bickenbach Key words: keratinocytes epidermis hair follicle bulge transit amplifying cells b 1 integrin, a 6 integrin gene therapy In recent years, cutaneous epithelial stem cells have attained a genuine celebrity status. They are considered the key resource for epidermal and skin appendage regeneration, and are proposed as a preferential target of cutaneous gene therapy. Follicular epithelial stem cells may also give rise to a large variety of epithelial tumors, and cutaneous epithelial stem cells likely are crucial targets for physical or chemical agents (including carcinogens) that damage the skin and its appendages. However, as this Controversies feature illustrates, few experts can agree on how exactly to define and identify these elusive cells, or on where precisely in the skin they are localized. Given their potential importance in skin biology, pathology and future dermatological therapy, it is, therefore, timely to carefully reconsider the basic questions: What exactly is a stem cell, and how can we reliably identify epithelial stem cells? How many different kinds are there, and how do they differ functionally? Where exactly in the skin epithelium is each of the putative stem cell subpopulations located, and can we selectively manipulate any of them? Viewpoint 1 What is a stem cell? In the context of a rapidly proliferating tissue such as the cutaneous epithelium, a stem cell can be defined as a relatively undifferentiated, multipotent, generally quiescent cell with a high proliferative potential that gives rise to both other stem cells and more proliferative transit amplifying (TA) cells. TA cells are responsible for the immediate replenishment of cells that are lost to the environment after terminal differentiation. Other stem cell attributes include the ability to proliferate in response to wounding and hyperproliferative stimuli, high b 1 integrin expression and a life span which exceeds that of the organism (for reviews, see (1, 2)). Although enough evidence exists to support the concept that the epidermis is organized into a hierarchy of stem, TA and terminally differentiated cells, in hair-bearing skin hair follicle bulge keratinocytes best satisfy the wide array of stem cell criteria (see Table 1). This suggests that, under physiological conditions, the hair follicle bulge serves as the ultimate reserve of both epidermal and hair follicle keratinocytes. How are stem cells identified? In vivo detection of slowly-cycling cells The two most salient features of stem cells, namely their slowly-cycling nature and high proliferative potential, can be identified by in vivo and in vitro 80

2 Controversies Table 1. Stem cell characteristics of hair follicle bulge cells compared to other possible stem cell populations Bulge cells Hair matrix cells Interfollicular Stem cell Central cell of EPU epidermal cells attributes Rodent Human Rodent Human Rodent Human slowly cycling yes (5) yes (28) no (5) no yes, but not as likely (16, 17) slow as bulge (6) high proliferative potential yes (13)?/not under current no, not under no, not under? yes (12) in vitro (CFE) culture conditions current culture current culture conditions (13) conditions (14) in vivo?? no most likely? yes (23) longevityωduration of lifetime likely (6)? no no no (6)? multipotent likely (20) likely (14) likely (24) likely no no express high integrin levels? yes (28)? no (28)? yes (15) proliferate at anagen onset yes (25) yes (28) no no NR NR distinct keratin profile yes (26) yes may not express? no no keratin? (24) proliferate in response yes (25)? NR NR yes likely to wounding NR: not relevant; EPU: epidermal proliferative unit; CFE: colony forming efficiency. methodologies, respectively. Label-retaining cell analysis results in detection of quiescent cells in vivo (3, 4). Presumably because stem cells are highly proliferative during development they can be labeled with 3 H-TdR or BrdU administration during the neonatal time period. After a chase period when no label is administered for 4 6 weeks, only slowly-cycling cells remain labeled and are detected with autoradiography or immunohistology. These types of studies in mice clearly demonstrate that label-retaining cells are present in both the epidermis and hair follicle bulge (3 5). However, recent studies utilizing very long chase periods for over 1 year show that the slowest-cycling cells within the entire cutaneous epithelium reside in the bulge (6). Given that the life span of a mouse is roughly 2 years, it is likely that these cells persist throughout the mouse s lifetime. In fully developed animals, stem cells are more difficult to label, because they proliferate less frequently, and conventional wisdom states that only prolonged continuous labeling will label stem cells in adult animals (7, 8). Similarly, it is thought that stem cells can not be labeled with a pulse of 3 H- TdR (9, 10). However, important exceptions exist to both of these statements. Even a single pulse of BrdU labels stem cells if they are proliferating when the pulse is delivered. For example, at the onset of a new hair growth cycle (anagen), a pulse of BrdU results in label-retaining cells that persist for at least several weeks (personal observation). Conversely, even a 2-week labeling period will not label hair follicle stem cells if the follicle does not traverse through anagen onset (when stem cells are proliferating). Therefore, prolonged labeling periods do not guarantee that stem cells will be detected as label-retaining cells, and it is even more precarious to suggest that all stem cells are detected by these techniques. These concerns are par- Figure 1. Epithelial stem cell system of the cutaneous epithelium. Epithelial stem cells in the hair follicle bulge give rise to at least 4 progenitor cell types (1 epidermal and 3 hair [see ref. (20)]) DP: dermal papilla; EPC: epidermal progenitor cell; EPU: epidermal proliferative unit (27) HPC: hair progenitor cell; SC: stem cell. 81

3 Cotsarelis et al. ticularly relevant to recent studies in which a 1- week labeling period with BrdU was utilized to detect stem cells in the corneal epithelium (11). The investigators assumed that all stem cells were labeled after 1 week of continuous BrdU administration in adult animals, yet typically only 80% of limbal cells are labeled after 2-weeks of continuous labeling (9), suggesting that many cells in the limbus do not divide in a 2-week period. In vitro assessment of proliferative potential Individual epidermal keratinocytes display varying abilities to form colonies in cell culture systems (12). This ability is thought to correlate to the proliferative potential of a cell. The number of keratinocyte colonies generated from various portions of the hair follicle can be counted and colony forming efficiency (CFE) can be calculated as (number of colonies)/(total number of plated cells) (13). This value can be used as an indication of the number of stem cells in a particular region of the epithelium. In rats, this type of analysis localizes stem cells to the hair follicle bulge (13). Surprisingly, in human hair follicles, colony forming efficiency localizes stem cells to outside of the bulge region, in the lower outer root sheath (14). This portion of the follicle undergoes degeneration during hair follicle regression (catagen), therefore the exact location of human hair follicle stem cells is unclear. More suitable culture conditions for hair follicle bulge and matrix cells are probably needed to accurately assess the proliferative potential of keratinocytes within the follicle. Current cell culture conditions may preferentially select for epidermal progenitor cells (see Fig. 1). Markers for epithelial stem cells Keratinocytes with high colony forming efficiency also possess high levels of b 1 integrin, therefore b 1 integrin is thought to be a stem cell marker (15). Integrin expression, in combination with other markers, may ultimately be useful for analyzing the precise location of stem cells, and for enriching these cells using cell sorting techniques (16, 17). Cytokeratins 15 and 19 may also be epithelial stem cell markers (26, 28). The hair follicle as a paradigm for studying stem cells The hair follicle contains rapidly proliferating and differentiating cells similar to the epidermis. Unlike the epidermis, however, hair follicle growth (anagen) is interrupted by periods of re- gression (catagen) and rest (telogen) (18, 19). The hair follicle produces at least 6 different types of keratinocytes probably from 3 distinct progenitor cells, as suggested by lineage studies (20). Since the slowest cycling keratinocytes are in the bulge, it is intriguing to postulate that it generates at least 4 different types of precursors (epidermal, outer root sheath, inner root sheath and hair shaft; see Fig. 1). Future studies need to identify the precise location of stem cells in human tissues. Consideration of stem cell proliferative behavior will be necessary for the success of cutaneous gene therapy, since expression and integration of transgenes is generally dependent on proliferation of host cells. With the recent advent of methodologies resulting in replacement of mutant genes with wild type sequences through homologous recombination (21, 22), permanent genotypic alterations in the cutaneous epithelium could be achieved by targeting epithelial stem cells. George Cotsarelis M8 Stellar-Chance Laboratories 422 Curie Blvd Philadelphia, PA cotsarel/mail.med.upenn.edu 1. Miller S J et al. Sem Dev Biol 1993: 4: Morrison S J et al. Cell 1997: 88: Bickenbach J R. J Dent Res 1981: Morris R et al. J Invest Dermatol 1985: 84: Cotsarelis G et al. Cell 1990: 61: Morris R J, Potten C S. Inv Dermatol 1995: 104: 578 (abstract). 7. Lavker R M et al. J Invest Dermatol 1993: 101: 16S 26S. 8. Miller S J et al. J Invest Dermatol 1993: 100: 288S 294S. 9. Cotsarelis G et al. Cell 1989: 57: Sun T T et al. J Invest Dermatol 1991: 96: 77s 78s. 11. Lehrer M S et al. J Cell Sci 1998: 111: Barrandon Y, Green H. Proc Natl Acad Sci USA 1987: 84: Kobayashi K et al. Proc Nat Acad Sci USA 1993: 90: Rochat A et al. Cell 1994: 76: Jones P H, Watt F M. Cell 1993: 73: Li A et al. Proc Natl Acad Sci USA 1998: 95: Batacsorgo Z et al. J Exp Med 1993: 0178: Cotsarelis G. Am J Pathol 1997: 151: Paus R. Curr Opin Dermatol 1996: 3: Kamimura J et al. J Invest Dermatol 1997: 109: Russell D W, Hirata R K. Nat Gen 1998: 18: Yoon K et al. Proc Nat Acad Sci USA 1996: 93: Compton CCetal. LabInvest1989: 60: Reynolds A J, Jahoda C A. Development 1996: 122: Wilson C et al. Differentiation 1994: 55: Michel M et al. J Cell Sci 1996: 109: Potten C S. Int Rev Cytol 1981: 69: Lyle et al. J Cell Sci 1998: 111:

4 Viewpoint 2 Controversies Are there epidermal stem cells? Stratified squamous epithelia, such as the epidermis, are continuously renewing tissues with structures that are maintained by division of cells in the proliferative layer to replace cells in the outer layer that are sloughed into the environment. This mechanism of balancing the rate of cell division with the rate of cell loss is essential for epithelial homeostasis and must be maintained for life (1). Although many researchers have been working on this concept for decades, the mechanisms controlling the relationship between cell division and cell differentiation are still not clear. In the mid 1960s, Leblond (2) theorized that cell division and cell migration were random events, that when a basal cell divided it squeezed another basal cell into the suprabasal layer. Later it was suggested that stratified squamous epithelia, similar to the hematopoietic system, consisted of a hierarchy of dividing cells maintained by a small subpopulation of stem cells (3). It was proposed that stem cells were a self-renewing population in which each stem cell division produced 1 stem cell to maintain the population and 1 transient amplifying daughter cell, which divided a few times but was ultimately committed to differentiation, whereas the stem cells remained in the basal layer for the lifetime of the epithelia (4). Although there is no direct evidence for differential stem cell division, this is still the belief held today. There is evidence for the existence of epidermal stem cells: 1) Clones formed from radioresistant epidermal cells repopulate the skin after severe radiation depletion (5, 6). 2) Allogeneic grafts form epidermal basal keratinocytes, expanded at least 10-fold in culture, have lasted for years on burn patients (7, 8). These grafts reform a fullystructured epidermis, although no hair follicles or sweat glands are formed. 3) Heterogeneity in the basal cell cycle was discovered, in which a small subpopulation of cells was shown to retain a tritiated thymidine label for up to 240 days (9), suggesting a very long cell cycle. These label-retaining cells (LRCs) are small, contain few organelles, occupy a fixed position in the tissue architecture, and are clonogenic in vitro. All are properties anticipated for epithelial stem cells. Thus, in the last decade, we have progressed from asking whether stem cells exist to asking: how many are there, and how can we separate them from other basal cells so that we can use them in gene therapy regimes? How many stem cells are there? Early reports on colony formation after epidermal irradiation (5, 6) suggested that the percentage of cells capable of regenerating a radiation-depleted epidermis ranged from 4 17%, depending on the radiation dose. Thus, a subpopulation of radioresistant cells exists in the epidermis, but it is not clear how many might be true stem cells. Approximately 40% of the basal cell population showed a bright expression of b 1 integrin (10), and it was hypothesized that these were the stem cells. However, since irradiation and cell kinetic analyses predicted that less than 10% of the basal cell population could be stem cells (5, 11), the b 1 -bright population must contain more than just stem cells, and thus cannot be used to predict the size of the population. It has been assumed that the number of epithelial stem cells is directly related to tissue architecture with 1 stem cell maintaining 1 unit of structure (12). The most supportive evidence for this theory is supplied in 2 recent articles, in which retroviral transduced epidermal cells appeared to produce recognizable columnar units of structure in grafts (13, 14). These data suggest that in mouse and human skin approximately 10% of the basal cells are stem cells. However, in a similar experiment using porcine epidermal cells, it was reported that a single stem cell was responsible for maintaining more than 1 epidermal papilla (15), which includes between 20 and 30 basal cells. This suggests that 3 4% of the basal cell population are stem cells. In another study, 10% of the basal cells rapidly adhered to collagen type IV, but only 40% of these cells formed large clones, suggesting that the rapidly adhering population also contains transient amplifying cells, and that only 4% of the basal cells are stem cells (16). Thus, the exact number of epidermal stem cells has yet to be clearly defined. Can stem cells be separated from transit amplifying cells? Since no specific cell surface markers that differentiate epidermal stem cells from transit amplifying cells have been identified, separation of pure live stem cells is not yet possible. However, enrichment for stem cells has been achieved using either specific antibodies or adhesion to ligands. In the bulge of hair follicles, LRCs, the putative stem cells, were detected by an antibody to K19, a keratin found 83

5 Cotsarelis et al. in simple epithelia, but no K19 staining was observed in LRCs in the interfollicular epidermis and K 19 stained more than just the bulge LRCs (17). Nine years ago, monoclonal antibodies were developed for transit amplifying cells. These did not stain LRCs, but instead identified basal cells that showed early stages of maturation and differing proliferative potentials (18). Early in this decade, the intensity of b 1, a 2, or a 3 integrin staining was used to identify a subpopulation of basal cells, of which some showed high proliferative potential (10). Similarly, the combination of high expression of a 6 integrin and low expression of a proliferative associated cell surface marker resulted in a population highly enriched for proliferative cells (19), although not all of the cells showed extended proliferation. More recently it has been shown that basal cells with high proliferative potential could be enriched by rapid adhesion to a variety of substrates (16), which suggests that stem cells, and perhaps early transit amplifying cells, may be stickier than other basal cells, but that this adherent property is unlikely to be mediated through 1 specific cell surface adhesion marker. Stem cells and gene therapy Since it is believed that only the stem cells remain for the lifetime of the epidermis (4), they must be considered as the target cells in gene therapy regimes for genetically inherited skin diseases (20). Although several have introduced recombinant genes into epidermal cells, in most cases, gene expression lasted less than 4 weeks (21, 22). Only a few reported longer expression and even then it was in less than 1% of the basal cell population (13, 14, 23), which suggests that very few stem cells were transfected. To introduce recombinant genes into epidermal stem cells one must either achieve 100% transfection efficiency in a population of total basal cells which contains a few stem cells, or separate the stem cells from the transit amplifying cells before transfection. The first is most commonly used, but this method appears to transfect very few stem cells. The second is more difficult since pure separation of epidermal stem cells has yet to be achieved. However, 2 recent reports have shown that substantial enrichment of stem cells before transfection is possible (16, 24). As the enrichment procedures improve, pre-selection is likely to be the first step in future gene therapy strategies for genetic skin diseases. Questions remaining to be answered Despite the extensive experimentation of the last 2 decades, the main questions concerning epider- mal stem cells still remain unanswered today. Since we still have no specific marker for stem cells, we don t know how many there are. We assume that stem cells are responsible for maintaining a unit of tissue structure, but we don t know the size of the structural unit or how many transit amplifying cells are involved. We aren t sure that transit amplifying cells cannot revert to stem cells after severe tissue damage, which brings me to my final point we don t even know whether being a stem cell is an intrinsic characteristic or whether any cell can respond to extrinsic factors and become a stem cell. With an increasing number of gene therapy regimes likely to depend on gene expression in epidermal stem cells, it is imperative to ascertain the answers to these questions. Jackie R. Bickenbach Dept of Anatomy & Cell Biology University of Iowa, Iowa City, PA jackie-bickenbach/uiowa.edn 1. Potten C S. Int Rev Cytol 1981: 69: Leblond C P et al. Adv Biol Skin 1964: 5: Cairnie A B et al. Stem Cells of Renewing Cell Populations. New York: Academic Press Cairns J. Nature 1975: 255: Withers H R. Radiat Res 1967: 32: Potten C S, Hendry H H. Int J Radiat Biol 1973: 24: Rheinwald J G, Green H. Cell 1975: 6: Compton CCetal. LabInvest1989: 60: Bickenbach J R. J Dent Res 1981: 122C: Jones P H, Watt F M. Cell 1993: 73: Clausen O P F, Potten C S. Cutan Pathol 1990: 17: Potten C S. Cell Tissue Kinet 1974: 7: Mackenzie I C. J Invest Dermatol 1997: 109: Kolodka T M et al. Proc Natl Acad Sci USA 1998: 95: Ng R L H et al. J Invest Dermatol 1997: 108: Bickenbach J R, Chism E. Selection and extended growth of murine epidermal stem cells in culture. Exp Cell Res 1998 (in press). 17. Michel M et al. J Cell Sci 1996: 109: Mackenzie I C et al. Differentiation 1989: 41: Li A et al. Proc Natl Acad Sci USA 1998: 95: Blau H, Khavari P. Nature Med 1997: 3: Morgan J R et al. Science 1987: 237: Fenjives E S et al. J Invest Dermatol 1996: 106: FlowersMEDetal.ProcNatl Acad Sci USA 1990: 87: Bickenbach J R, Roop D R. Transfection of a preselected population of human epidermal stem cells: consequences for gene therapy. Proc Assoc Amer Physicians 1998 (in press). 84

6 Commentary 1 Controversies Epithelial stem cells are indeed a controversial topic of research and nowhere is there a clearer indication of the personal bias of scientific investigators than when discussing these elusive cells. The problem begins in agreeing on the best definition of stem cells (SCs). Reading between the lines of Viewpoint 1, it appears that the hair follicle contains the ultimate SC population, given that the slowest cycling cells reside in the bulge region and that these cells also give rise to interfollicular (IF) epidermal cells under homeostatic conditions. There is no evidence to support this contention on the contrary, the IF epidermis is a self-renewing tissue containing its own slow-cycling SCs and transit amplifying (TA) cells (1 3; Viewpoint 2). The absence of label-retaining cells in the IF epidermis after prolonged periods may merely be a reflection of the differing cell kinetics and hierarchies within the two SC populations. To play devil s advocate, one might pose the question: what is the contribution of label-retaining cells that persist in the hair follicle for over 1 year to cell renewal? Indeed, can we be sure that they are epithelial cells? Extensive proliferative potential over the long term is another hallmark of SCs in vivo the confusion arises when we try to determine this potential in vitro. Murine keratinocytes are notoriously difficult to propagate in vitro, and it has been elegantly demonstrated that only label-retaining cells (SCs) were clonogenic in vitro, whilst pulselabelled cells (TAs) did not form colonies (4). In contrast, since human keratinocytes can be propagated in culture relatively easily and cells of varying proliferative potential can form colonies, investigators have assumed that SCs can be distinguished from TAs by their ability to form larger colonies and greater numbers of them (5, 6). These assumptions may be questionable, given that haematopoietic SCs with marrow repopulating activity have been shown to not clone directly in vitro, although they will over time in culture give rise to directly clonogenic cells (7 9). The most rigorous analysis of the SC population then must fulfil the greatest number of attributes assigned to these cells by careful kinetic studies in vivo (1 3). Neither Viewpoints adequately discusses a recent study from my own laboratory which has demonstrated the isolation of a putative SC population of human epidermis, which fulfils several such attributes. Two cell surface markers, i.e. a 6 integrin and a proliferation-associated cell surface marker termed 10G7 antigen, were utilized to select cells with the phenotype a 6bri 10G7 dim (10). These cells comprised a minor subpopulation of immature basal cells ( 10%), exhibiting the greatest regenerative capacity of all basal cells in long-term culture (80? days), and importantly were quiescent at the time of isolation from the epidermis. In addition, 2 other populations of basal cells were also identified: an actively cycling, putative TA fraction (a 6 bri 10G7 bri ) which showed lower long-term proliferative output; and a K10 positive post-mitotic differentiating fraction (a 6 dim ). Notably, the short-term colony forming ability of these TA and SC were not markedly different, lending weight to the argument that it may not be possible to distinguish these 2 populations of basal cells with this assay. Let us be very clear when we use the term stem cell marker to me this suggests a marker expressed exclusively by stem cells. In this light, it is true that there is no single stem cell marker (Viewpoint 2), even for haematopoietic SCs which are isolated on the basis of multiple composite cell surface phenotypes. Thus, given that all basal cells express integrins (both a 6 and b 1 ), it is quite inaccurate to suggest that b 1 may be a stem cell marker (cf. Viewpoint 1). This brings me to the much sought-after goal of localizing human epidermal SCs in vivo this is unlikely to be achieved in the absence of specific markers for this population. It has been suggested that rapidly adhering epidermal cells comprise epidermal stem cells (6). This is simply not the case. Recent data with human SCs and TAs (11), and those reported in Viewpoint 2 for murine LRCs, demonstrate that both SCs and TAs adhere rapidly to collagen type IV. Unfortunately for stem cell biologists, these data demonstrate that enrichment for SCs cannot be achieved by simply selecting for stickier cells. In conclusion, it is worth noting that unlike the haematologists, we are a long way from demonstrating that any putative epidermal stem cell population has the tissue reconstituting ability demonstrable for haematopoietic stem cells. It is perhaps not surprising then, that we cannot agree on whether anyone has come close to identifying epidermal stem cells. Pritinder Kaur Division of Hematology Hanson Ctr. for Cancer Res Inst. for Med. and Vet. Sci. Adelaide, SA 5000, Australia pritinder. kaur/imvs.sa.gov.au 85

7 Cotsarelis et al. 1. Potten C S. In: Stem Cells: Their identification and characterization, Potten C S, ed. London: Churchill Livingstone 1983: Morris R J et al. J Invest Dermatol 1983: 84: MacKenzie I C, Bickenbach J R. Cell Tissue Res 1985: 242: Morris R J, Potten C S. Cell Prolif 1994: 27: Barrandon Y, Green H. Proc Natl Acad Sci USA 1987: 84: Jones P H, Watt F M. Cell 1993: 73: Sutherland H J et al. Proc Natl Acad Sci USA 1990: 87: Haylock D N et al. Blood 1992: 80: Haylock D N et al. Blood 1997: 90: Li A et al. Proc Natl Acad Sci 1998: 95: Li A, Kaur P. Adhesive properties of human basal epidermal cells: an analysis of keratinocyte stem cells, transit amplifying cells and post-mitotic differentiating cells. Manuscript submitted. Commentary 2 The concept of stem cells, which is now clearly defined (see Viewpoint 1 and 2) was first advanced for 2 rapidly renewing tissues: the epidermis and the hematopoietic system. In fact, each adult tissue, including even the CNS (1) may involve stem cells. In skin, despite all the studies concerning the epidermis, we must not preclude the possibility that stem cells exist in the dermis. The cutaneous epithelial stem cells have been isolated in vivo by microdissection (2) or in vitro through their differential adhesive properties (3, 4). Two major and related questions remain to be answered and represent an important challenge for the future. How many different types of stem cells exist in skin epithelia? Cutaneous epithelia are likely to contain pluripotent stem cells which can give rise to partially committed progenitor cells restricted along a specific lineage. Stem cells from a common ectodermal embryonic origin could be equipotent. Data from Dr Jahoda s group (5) and from my group (6) demon- Figure 1. During embryogenesis, a series of 4 binary choices leads to the formation of at least 5 types of progenitor cells in the integument and to the three locations of stem cells with apparently the same abilities. 86

8 Controversies strate that adult epidermal keratinocytes, even after in vitro culture, can be induced by dermal papilla cells to form hair follicles. Conversely, the outer root sheath cells can regenerate an epidermis (7). Moreover, rabbit corneal epithelium from 1-weekold offspring (8), or even from an adult (9), gives rise to an epidermis and to hair follicles when associated with an embryonic hair-forming mouse dermis. Moreover, even the late transient amplifying cells located in the center of the adult cornea (10) are able to revert to stem cells (9). In these experiments, the origin of cells was conclusively established by the different chromatin pattern between the 2 species. Thus, the stem cells which are located in 3 different areas of the integument, namely epidermal rete ridges, corneal limbus, and hair bulge (or a circumscribed section of the outer root sheath), appear to be identical or at least interchangeable (Fig. 1). In other words, only 1 type of cutaneous stem cells may exist, which are located in 3 different niches. The progenitor cells to which these stem cells give birth may depend on their respective mesenchymal environment. How are stem cells individualized? The patterning of stem cells within the epithelia could be generated by cell cell interactions among the keratinocytes. These interactions may depend on interactions with the dermis, generated by cell cell interactions among the fibroblasts. Recent results (11, 12) show that intercellular interactions, both in the epidermis and the dermis are mediated by Notch signalling, while dermal epidermal in- teractions, as shown by many authors, are mediated by growth factors. The mammalian proteins belonging to the Notch signalling pathway are homologues of Drosophila cell-fate patterning genes. These proteins appear to specify the binary choices (Fig. 1) which occur at each step of integument individualization. However, the interactions between the Notch system and selected cell cell adhesion molecules well known to be one of the characteristics (3, 4) of the stem cells remain to be elucidated. Danielle Dhouailly Biologie de la Différenciation Epithéliale LEDAC UMR CNRS 5538 Institut A. Bonniot La Tronche, France danielle.dhouailly/ujf-grenoble.fr 1. Reynolds B A, Weiss S. Science 255: Rochat A et al. Cell 1994: 76: Jones P H. Bioessays 1997: 19: Bagutti C et al. Dev Biol 1996: 179: Reynolds A J, Jahoda C A B. Development 1992: 115: Ferraris C et al. Int J Dev Biol 1997: 41: Limat A M, Noser F K. J Invest Dermatol 1986: 87: Ferraris C et al. Differentiation 1994: 57: Ferraris C et al. Transient corneal epithelial cells are able to revert to stem cells and to give rise to an epidermis and hairs or sweat glands, depending of the origin of the associated embryonic dermis (Submitted). 10. Lehrer M et al. J Cell Sci 1998: 111: Viallet J P et al. Mech Dev 1998: 72: Thelu J et al. J Invest Dermatol 1998: 111: Commentary 3 Since there is agreement that epithelial cells with stem cell-like behavior exist, various attempts to define their biochemical and functional properties have been undertaken. However, it is pertinent to rely on in vivo models for two reasons: only time will prove the persistence of stem cells that demonstrate the capacity to generate a fully mature epidermis. Secondly, stem cell-like behavior is a consequence of complex and permanent interrelations between basal keratinocytes and extracellular matrix, basal membrane, combinations of growth factors, dermal cells and nerve endings that cannot be simulated in cell or organ culture experiments (3, 5), which are not depicted in Dr Cotsarelis figure. As he indicated, cell culture experiments are useful to estimate the intrinsic growth potential of certain clones, but they may just reflect the ability of individual cells to cope with artificial and depriving conditions in a system prone to differentiation and death. Moreover, it is wise to assume a high degree of plasticity and permanent remodeling rather than a static hierarchy of keratinocytes. For example, isolated dermal papilla cells have been found to induce follicle formation and hair growth by transdifferentiation of adult rat epidermis (11). In addition, the activation 87

9 Cotsarelis et al. of the stromal microenvironment was shown to induce the formation of benign cystic tumors when HaCat cells expressed platelet-derived growth factor (PDGF) in vivo (12). Recent gene marking studies using retrovirusmediated gene transfer have revealed various clues to the clonogenic potential and persistence of transduced keratinocytes in vivo (7, 9). In these elegant studies convincing evidence has been generated in favor of distinct vertical columns of the epidermal proliferative unit as suggested by Potten (10). In the study by Kolodka et al. the fraction of stem cells that was transduced ranged between 21 and 28% and was thus considerably higher than stated by Dr Bickenbach. Why silencing of the transgene did not occur in this study, in contrast to many others, still has to be answered (1, 2). With constant progress in defining surface markers for stem cell-like behavior such as a 6 bright 10G7 dim it should become possible to transfer genes of interest into a highly enriched stem cell population leading to increased amounts of gene product, potentially reaching therapeutic levels (8). Since the epidermal stem cell marker has not yet been found, if it exists at all, stem cell characterization should also include indicators of metabolism, such as sorting by the paucity of mitochondria in slow-cycling cells and the abundance of endoplasmatic reticulum in transit amplifying cells using organelle-selective dyes. Some of the difficulties in distinguishing stem cells from transit amplifying cells arise from the dogmatic view of the unidirectionality of cell differentiation, as correctly pointed out by Dr Bickenbach. However, few attempts have been made to investigate whether a certain genetic program (e.g. programmed dedifferentiation or transdifferentiation) may be initiated under certain environmental conditions such as wounding, burn or hyperproliferative diseases (11). In this context, attention should be paid to certain homologues of Drosophila cell fate patterning genes such as Notch, Delta-1 and Wingless (4). Alternatively, stem cell behavior could also be under negative growth control, with differentiation beginning upon the alteration of intercellular (E-cadherin, plakoglobin and beta-catenin) and basement membrane (b 1 integrin) contacts. As to the number and residence of epithelial stem cells, essential features still have to be elucidated. The higher density of epidermal stem cells in the bulge area may simply reflect the increased proliferative activity in hair follicles (growth rate of 0.2 mm per day), compared to interfollicular epidermis. Furthermore, the regeneration of epidermal cells on the palms, soles, cornea and mucosa, which are renewed every 3 days, does not support the bulge model of stem cell residence (6). In addition, how do we explain the epidermal renewal in disease states of the hair follicle such as scarring alopecia and graft versus host disease of the integument? Ulrich Hengge Dept. of Dermatology University of Essen Hufelandstr. 55 D Essen Tel.: π dermatology/uni-essen.dl 1. Deng H et al. Nat Biotechnol 1997: 15: Fenjves F S et al. J Invest Dermatol 1996: 106: Fusenig N E. In: Leigh L et al., eds. The Keratinocyte Handbook. Cambridge: Cambridge University Press, 1994: Go M J et al. Development 1998: 125: Hall P A, Watt F M. Development 1989: 106: Hengge U R et al. Efficient expression of naked Plasmid DNA in mucosal epithelial prospective for the treatment of skin lesions. J Invest Dermatol (in press). 7. Kolodka T M et al. Proc Natl Acad Sci USA 1998: 95: Li A et al. Proc Natl Acad Sci USA 1998: 95: Ng R L H et al. J Invest Dermatol 1997: 108: Potten C S. Cell Tissue Kinet 1974: 7: Reynolds A J, Jahoda C A B. Development 1992: 115: Skobe M, Fusenig N F. Proc Natl Acad Sci USA 1998: 95: