keratinocyte stem cell assays：an evolving science

Keratinocyte Stem Cell Assays: An Evolving Science Pritinder Kaur,� Amy Li,� Richard Redvers,� and Ivan Bertoncellow �Epithelial Stem Cell Biology Laboratory; wHaemopoiesis Laboratory, Peter MacCallum Cancer Centre, Melbourne, Victoria, Australia Although the existence of epithelial stem cells in the skin has been known for some decades from cell kinetic studies performed in vivo, attempts to prospectively isolate these cells for further biological characterization have been made possible relatively recently facilitated by the availability of antibodies that detect cell surface markers on epidermal cells. Elegant gene marking studies in vivo have provided confirmation of the patterns of epithelial tissue replacement predicted by classical cell turnover studies. But, the identification of candidate epidermal stem cells ex vivo remains an area of great controversy, requiring the re-evaluation of current experimental approaches that rely of necessity on predicted epidermal stem cell behavior in culture. Here we review the diverse experimental approaches utilized to identify keratinocyte stem cells and their underlying assumptions. We conclude that hair follicles and interfollicular epidermis each have their own self-renewing stem cell populations, contributing to distinct regions of the epithelium during homeostasis, although this is perturbed during wound healing. The need for the development of more rigorous assays for stem cell activity is highlighted given our recent observations using current assays and the discovery of new surface markers that identify putative epidermal stem cells. Key words: Self-renewal/in vitro/transplantation/methods/keratinocyte stem cells J Investig Dermatol Symp Proc 9:238 –247, 2004 The extensive capacity of the epidermis for cell renewal in vivo has been demonstrated by their ability to survive, expand, and generate cultured human epidermal sheets capable of rescuing patients with full thickness burns covering up to 98% of their body surface that are maintained for over a decade (Gallico et al, 1984; Compton et al, 1989). Al- though these studies demonstrate the immense regenera- tive capacity of keratinocytes, the stem cells in this tissue remain uncharacterized. Extensive investigation combining close histological analyses and elegant in vivo cell kinetic studies conducted worldwide some 25 y ago firmly estab- lished that murine interfollicular (IF) epidermis is a highly organized, stratified tissue with an impressive cell turnover rate which is ultimately dependent on a minor population of resident stem cells (Mackenzie, 1970; Allen and Potten, 1974; Potten, 1983; Morris et al, 1985). Thus, mature cells or squames lost from the skin surface are continuously re- placed by a carefully orchestrated process of cell prolifer- ation within the basal layer, which lies adjacent to the basement membrane. Similar in vivo cell kinetic analysis in mice subsequently established the presence and role of epithelial stem cells in cell renewal at other anatomical sites including the bulge region of the hair follicle (HF), limbus of the cornea, the basal layer of oral mucosa, and in the base of small intestinal crypts (reviewed by Miller et al (1993)). The development of cell culture techniques permitting the in vitro propagation of epithelial cells was a further critical step advancing the study of epithelial stem cells (Rheinwald and Green, 1975) providing the impetus for skin biologists to identify, isolate and assay epidermal stem cells from explanted tissue using the experimental approaches and criteria developed in dissecting and ordering the hemo- poietic stem and progenitor cell hierarchy as a template. Although the functional attributes of epithelial stem cells and hemopoietic stem cells will be different, reflecting their diverse biological roles in the body, they share many op- erational characteristics of adult stem cells of all conti- nuously renewing cell populations: low incidence, low probability of cycling, slow turnover rate, the ability to self-renew, and the ability to regenerate and repair tissue in the steady state and following damage. Thus, colony-form- ing assays, alone or in combination with fluorescence ac- tivated cell sorting (FACS), and gene marking combined with the analysis of tissue regeneration in vitro or in vivo (Barrandon and Green, 1987; Jones and Watt, 1993; Jones et al, 1995; Mackenzie, 1997; Kolodka et al, 1998; Li et al, 1998; Tani et al, 2000; Ghazizadeh et al, 2001) have proved extremely useful in beginning to extend our knowledge of epidermal stem cells. As in the hemopoietic system, the continuing challenge for epithelial stem cell biologists is the development and refinement of predictive surrogate assays which provide a measure of stem cell activity in cell populations removed from their native microenvironment. The controversy cur- rently hotly debated in scientific forums is almost entirely cen- tered around the validity of specific molecular markers and assays employed to identify epidermal stem cells either in situ or ex vivo, and the validity of assays which purportedly distinguish stem cells from their more committed progeny. In the absence of (a) a rigorous epidermal stem cell assay analogous to bone marrow reconstitution with candidate Abbreviations: EPU, epidermal proliferative unit; FACS, fluores- cence activated cell sorting; HF, hair follicle; IF, interfollicular; KSC, keratinocyte stem cell; LRC, label-retaining cell; PMD, post-mitotic differentiating; SP, side population; TA, transient-amplifying Copyright r 2004 by The Society for Investigative Dermatology, Inc. 238 hemopoietic stem cells in lethally irradiated mice; (b) a unique phenotypic marker repertoire; and (c) recombinant cytokines that have enabled hemopoietic stem cell biolo- gists to hierarchically order closely related marrow stem and progenitor cell populations by in vitro surrogate assays (Bertoncello and Bradford, 1997), epithelial stem cell biol- ogists have had to resort to readouts which are largely based on untested assumptions about the expected prop- erties and behavior of isolated epidermal or keratinocyte stem cells (KSC). Consequently, individual laboratories have favored the use of different surrogate assays which they feel best measure the characteristic(s) they regard as their gold- standard criteria defining a stem cell. These include, colony- forming efficiency, colony size and morphology, epithelial regeneration, and long-term proliferative potential in vitro. In this review, we examine the evidence and assumptions supporting the various experimental approaches taken to date in evaluating epidermal stem cell potential in vivo and in vitro and discuss their limitations. Evidence Supporting the Presence of Stem Cells in the IF Epidermis In Vivo The dorsal epithelium in mice is a complex tissue special- ized into HF, sebaceous glands, and the IF epithelium, each characterized by its own distinct program of differentiation. Recent developments have led to re-examination of the long-held view put forward by Potten, Mackenzie, Bicken- bach, and Morris that the IF epidermis is a self-renewing tissue even in hairy skin. These investigators generated a vast body of compelling cell kinetic data which dem- onstrated unequivocally that cell replacement in the IF epidermis occurs within small packets of self-renewing ep- idermis termed the epidermal proliferative unit or EPU com- prising about ten basal cells and their suprabasal progeny lying directly above them. Three classes of basal keratin- ocytes have been identified by cell turnover studies and painstaking spatial analysis in situ. The long-lived KSC comprise a minor subpopulation ( � 1%–10% of basal cells) within the center of the EPU that are relatively quiescent, and identified as single label-re- taining cells (LRC), after a prolonged chase period (6–8 wk) following repeated administration of 3H-Tdr. Short-lived transient-amplifying (TA) cells ( � 60% of basal cells) locat- ed peripherally to the KSC are rapidly proliferating cells that readily incorporate 3H-Tdr, but are lost from the basal layer to terminal differentiation within 4–5 d (Mackenzie, 1970; Potten, 1974; MacKenzie and Bickenbach, 1985; Morris et al, 1985; Bickenbach et al, 1986; see Miller et al (1993) for detailed review). A third class of basal post-mitotic differ- entiating (PMD) keratinocytes located at the edges of the EPU, exhibit the early stages of keratinization as judged by morphology and ultrastructure (Christophers, 1971; Allen and Potten, 1974). These PMD cells retain some contact with the basement membrane but have a shape suggestive of their being in the process of migrating out of the basal layer. They can be visualized as K10-positive ‘‘hand-mirror’’- shaped cells found in both murine and human epidermis (Schweizer et al, 1984; Mackenzie et al, 1989; Kaur and Li, 2000). It is well accepted that these three subsets of murine basal epidermal cells exist in vivo at specific anatomical sites, i.e., the IF epidermis of the dorsum, ear, and tail. The interrelationship of these subsets (i.e., KSC ! TA ! PMD cells) is also apparent and a reasonable model for cell re- newal. Thus KSC, defined here as LRC generated in new- born mice, are long-lived and persist into adulthood, suggesting that they are permanent residents in these tis- sues. TA cells have a short lifespan in vivo and are lost to terminal differentiation within weeks. And, migration studies tracking the progress of 3H-Tdr-labelled cells from the basal layer into suprabasal layers, provide strong evidence for the maturation of TA cells into basal PMD cells, and then ter- minally differentiating suprabasal cells. Importantly, the sus- pected progression of KSC ! TA ! PMD cells has recently been confirmed by elegant gene marking and lineage anal- ysis studies in two independent laboratories. Thus, cultured murine and human keratinocytes comprised of a mixture of b-galactosidase transduced and untransduced cells, trans- planted in vivo, regenerated a chimeric epithelium with interspersed columns/EPU of transduced (blue) cells (Mac- kenzie, 1997; Kolodka et al, 1998), providing strong evi- dence for the clonal derivation of the epidermis from IF stem cells as originally proposed by Mackenzie and Potten three decades ago (Mackenzie, 1970; Allen and Potten, 1974). Most recently, lineage analysis studies of epithelial cells re- trovirally marked in situ have provided compelling evidence supporting the self-sustaining units of epidermal cell re- newal or EPU even in hairy dorsal skin followed over 33 wk (Ghazizadeh et al, 2001). These data collectively point to a central role for resident IF stem cells in homeostatic cell renewal and the generation of mature keratinocytes for the lifetime of a mouse in vivo. Physiological Role of HF Stem Cells Kinetic studies have revealed the presence of stem cells visualized as slow-cycling cells/LRC in the bulge region of HF as well as the limbal region of the cornea and the dorsal tongue mucosa of adult mice (Bickenbach, 1981; Morris et al, 1985; Bickenbach et al, 1986; Cotsarelis et al, 1989, 1990; Lavker et al, 1991; see Miller et al (1993) for detailed review). The central role of the bulge region in cell renewal is supported by studies of human HF demonstrating that epi- thelial cells within this region have greater proliferative po- tential than that of IF epidermal cells in culture (Yang et al, 1993; Rochat et al, 1994). A number of recent observations have led to the concept that the HF stem cells found in the bulge region are the ultimate precursors of the IF epidermis under homeostatic conditions, and further that the IF ep- idermis regenerating cells can be viewed as TA cells (Lavker and Sun, 2000; Taylor et al, 2000). Specifically, a number of exquisite in vivo studies tracking the migration of b-gala- ctosidase- or BrDU-marked cells out of the bulge region have demonstrated the unequivocal role of these stem cells to give rise to all HF cell lineages, the sebaceous gland, and importantly to those regions of the IF epidermis adjacent to the HF. Whereas the contribution of follicle-derived cells to KERATINOCYTE STEM CELL ASSAYS 2399 : 3 SEPTEMBER 2004 IF epidermis was seen after implanting marked cells into wounded skin (Oshima et al, 2001) or in neonatal skin (Taylor et al, 2000), the work of Ghazizadeh et al (2001) clearly addressed the contribution of adult follicular cells to the IF epidermis under physiological conditions, i.e., up to 33-wk post-wounding. The latter study concurred that IF epider- mal cells adjacent to HF originated from follicular stem cells. But importantly, self-sustaining units of epidermal cells not associated with HF were consistently observed (Ghaziza- deh et al, 2001). Collectively, these studies demonstrate that the HF is an important reservoir for emergency repopulation of the IF epidermis following wounding and in the neonate as suggested previously (Eisen et al, 1956; Krawczyk, 1971; Al-Barwari and Potten, 1976; Cotsarelis et al, 1990). Further, parts of the IF epidermis are clearly derived from the HF routinely, but there is a class of IF epidermal stem cells that are capable of self-renewal throughout the lifetime of the animal, independent of the HF. Another confounding factor in the HF versus IF stem cell debate has been the observation that some of the slowest cycling cells, i.e., LRC, are readily found in the bulge region and not in the IF epidermis. It is important to bear in mind that IF stem cells exist as single cells at the center of EPUs and it is therefore harder to locate these single cells in a sea of unlabelled cells in randomly cut sections. In contrast, the follicular stem cells exist in a geographically distinct site in clusters of slowly cycling cells. Further, in common with the hemopoietic system (Bradford et al, 1997), it is critical to note that the epithelial stem cell compartment is a dynamic population of proliferative cells that cycle at a reasonable rate, although clearly less frequently than TA cells. Thus, the numbers of LRC found in any tissue at a given time is likely to reflect the number of divisions it has undergone since being labelled, the turnover rate of that tissue, and therefore the demands on the LRC for cell regeneration. That is to say, DNA label retention is a dynamic property. A study by Bickenbach quantitating the steady decline in numbers of LRC at different rates in different epithelia over time ele- gantly illustrates this point (Bickenbach et al, 1986). Thus, LRC have to be evaluated in the context of the time over which label retention is estimated. The difficulty in finding LRC in the IF epidermis after 8–10-wk post-labelling then, can most likely be attributed to the rate at which these cells have to divide to maintain the epidermis, resulting in a more rapid loss of label and that they are much harder to find. So, how slowly does a cell have to cycle to qualify as a stem cell? The point is that one cannot compete for stem cell status based on the rate at which a DNA label is retained because this merely reflects the cellular kinetics of a par- ticular epithelium. In fact, if a cell retains its DNA label for a prolonged period of time, such as the single LRC found in the bulge region a year post-labelling (Morris and Potten, 1999), one could argue that it has contributed little to cell regeneration. In Vitro Approaches to Defining Epithelial Stem Cells It can be confidently stated that all attempts to study ep- idermal stem cells in vitro are based on very specific as- sumptions about the intrinsic properties of stem cells made by the many investigators who have dared to venture into this minefield of experimental biology. But it should also be noted that the context in which stem cells are assayed is an equally, if not more important determinant of the assay readout. Not only does the act of removing candidate stem cells from their microenvironmental niche potentially alter their behavior, but the failure of surrogate assays to reca- pitulate the conditions under which stem cells grow and function in situ exerts artificial proliferative and differentia- tive pressures on these cells which will irreversibly alter their properties and influence their fate (Lavker and Sun, 2000). This philosophy is best encapsulated in the following state- ment made by Potten: ‘‘Stem cells are defined by virtue of their functional attributes. This immediately imposes diffi- culties since in order to identify whether a cell is a stem cell or not its function has to be tested. This inevitably demands that the cell must be manipulated experimentally, which may alter its properties’’ (Potten and Loeffler, 1990). In sup- port of this, Morris and Potten (1994) have elegantly dem- onstrated that murine epidermal 3H-Tdr LRC (putative KSC) can be recruited into proliferation upon placing them into culture. Thus clearly, stem cells do not remain quiescent when cultured; however, an unmistakable assumption in some of the pioneering work in the epidermal stem cell area is the notion that other subpopulations of keratinocyte pro- genitors, namely TA cells and differentiating cells, exhibit similar restricted proliferative potential when explanted in vitro as they do in vivo. For example, Barrandon and Green (1987) placed individual human keratinocytes in culture and could subsequently identify three classes of clonal cells: holoclones, which gave rise to large, rapidly growing col- onies; paraclones, which formed small colonies that under- went terminal differentiation after a few cell divisions; and meroclones, which gave rise to a mixture of growing and abortive colonies. These workers concluded that holo- clones were likely to be founded by KSC and paraclones by TA cells. Importantly, this interpretation was based on the assumptions that: (a) TA cells when explanted into culture would not be able to proliferate extensively in vitro and (b) analogous to the hemopoietic colonies with high prolifera- tive potential, the large keratinocyte colonies (holoclones) must be founded by stem cells. It remains unclear exactly which class of keratinocyte progenitor is represented by meroclones. In favor of the interpretation that holoclones are at least enriched for stem cells is the work from De Luca and Pellegrini showing that holoclones are capable of ex- tensive long-term proliferation measured over many weeks in culture both in skin (Mathor et al, 1996) and in the cornea (Pellegrini et al, 1999). Importantly, the limbal region of the cornea known to be enriched for LRC gives rise to holo- clones, whereas the central cornea does not (Pellegrini et al, 1999). An equally reasonable interpretation of these observa- tions could be that holoclones arise from both stem and TA cells, meroclones arise from more committed TA cells and that paraclones arise from keratinocytes that have initiated differentiation. Alternately, it could be argued that dispersing keratinocytes and placing them in culture perturbs the system so greatly that their behavior cannot reasonably be equated to that of any specific keratinocyte progenitor 240 KAUR ET AL JID SYMPOSIUM PROCEEDINGS in vivo under homeostatic conditions, or as stated by Bar- randon and Green (1987), ‘‘the relation, if any, between the clonal types we have described and the multiple cell types defined by thymidine labelling kinetics remains to be clarified’’. Identification of Markers for the Detection of KSC Watt and colleagues described the first attempt to pro- spectively define and isolate IF epidermal stem cells using cell surface markers and FACS techniques, an important prerequisite for further characterization of these cells that had best been studied in situ until then. Jones and Watt (1993) demonstrated that cultured human foreskin keratin- ocytes expressing high levels of b1 integrin had a hig