Keratinocyte stem cells: Friends and foes

Skin and its appendages provide a protective barrier against the assaults of the environment. To perform its role, epidermis undergoes an ongoing renewal through a balance of proliferation and differentiation/apoptosis called homeostasis. Keratinocyte stem cells reside in a special microenvironment called niche in basal epidermis, adult hair follicle, and sebaceous glands. While a definite marker has yet to be detected, data raised part in humans and part in the mouse system point to a critical role of stem and its progeny transit amplifying cells in epidermal homeostasis. Stem cells are protected from apoptosis and are long resident in adult epidermis. This renders them more prone to be the origin of skin cancer. In this review, we will outline the main features of adult stem cells in mouse and humans and discuss their fate in relation to differentiation, apoptosis, and cancer. J. Cell. Physiol. 225: 310–315, 2010. © 2010 Wiley‐Liss, Inc.

Epidermis serves as a barrier that protects against environmental stresses, such as harmful microbes, water loss, physical, thermal, and mechanical injuries. To support this role, the epidermis generates a number of appendages, including nails, hair follicles, sebaceous, and sweat glands. Epidermis must renew constantly throughout life to maintain normal homeostasis and to repair damage after wounding. The regenerative capacities of epithelial populations have been studied for many years with the development of cultured keratinocytes derived from the epidermis, used to produce autologous grafts that regenerate an epidermis over a full-thickness wound (Coolen et al., 2007). Epidermal homeostasis depends on a balance between proliferation and differentiation/apoptosis of keratinocytes. The epidermis contains a basal layer of proliferative keratinocytes that adhere to the underlying basement membrane (BM). Periodically, these cells detach from the BM, withdraw from the cell cycle, and initiate a program of terminal differentiation, moving upward to the skin surface. Upon commitment to terminal differentiation, keratinocytes progress through different stages and form three distinct layers: the spinous, the granular, and the stratum corneum. Eventually, keratinocytes loose the nuclei and are continuously shed from the skin surface. Although the morphogenetic changes associated with epidermal differentiation are well established, the molecular mechanisms underlying this process are complex, as they involve many genes and they are still poorly understood (Blanpain and Fuchs, 2009).
Keratinocyte stem cells (KSCs) are responsible for maintaining epidermal homeostasis and for repairing the tissue following injuries. KSCs self-renew and generate the differentiated lineages that form the mature tissue. There are three pools of KSCs in different locations. The interfollicular (IF) stem cells of the basal layer, the hair follicle (HF) stem cells of the bulge, and the sebaceous gland (SG) stem cells that are located above the bulge and below the hair shaft orifice (Fig. 1). During physiological tissue renewal, the three skin cell populations are each believed to be maintained by their own stem cells. When tissue homeostasis is disrupted, however, any of the three stem cell populations is capable of producing all three structures (Levy et al., 2007;Fuchs and Horsley, 2008). Human epidermis is thick with frequent cellular turnover and a tendency to produce long-term epidermal cultures in vitro. On the other hand, while the mouse can be genetically manipulated, it displays a thin, less active epidermis, whose cells fail to yield long-term cultures. In addition, some of the markers used to identify stem cells are different in mouse versus human skin. Despite these differences, common signaling pathways appear to control epithelial stem cell maintenance, activation, lineage determination, and differentiation. Data and discussion reported in this review will be based on studies performed in both mouse and human epidermis (Table 1).

Interfollicular Stem Cells
Regeneration of epidermis depends on the proliferation of a subpopulation of basal keratinocytes, known as IF stem cells. The current dogma of the epidermal proliferative unit (EPU) defines a model where a single stem cell divides infrequently to produce a stem cell daughter and a non-stem committed progenitor cell, named transit amplifying (TA) cell. TA cells divide a small number of times before withdrawing from the cell cycle and undergoing terminal differentiation in the suprabasal layer. The EPU model has been questioned recently by lineage tracing experiments demonstrating that normal adult epidermis is maintained by a single population of committed progenitor cells with different stochastic choices (Clayton et al., 2007). In any event, the EPU model implies that in the basal layer of epidermis there is a proliferative heterogeneity. Clonogenicity assays allow to distinguish three types of clones: holoclones that arise from stem cells and found large colonies with the greatest regenerative capacity in long-term culture; paraclones which give raise to abortive colonies and undergo differentiation; and meroclones that originate from TA and have an intermediate proliferative capacity (Barrandon and Green, 1987). The different clonogenic potential well correlates with the expression of b 1 integrin (Jones and Watt, 1993). Indeed, keratinocytes expressing highest levels of b 1 integrin give rise to holoclones, whereas cells with low levels of b 1 integrin produce smaller colonies. This has allowed to select a population of stem keratinocytes based on the rapid adherence to type IV collagen. A number of markers to enrich for stem cells have been reported, including high expression of a 6 integrin (a bri 6 ) and low expression of CD71 (CD71 dim ). In particular, a bri 6 CD71 dim basal keratinocytes exhibit many features of stem cells, such as quiescence and the greatest long-term regenerative capacity (Li et al., 1998). Furthermore, melanoma chondroitin sulfate proteoglycan (MCSP) is confined to non-cycling keratinocytes expressing highest levels of b 1 integrin, and can thus be considered a marker for epidermal stem cells (Legg et al., 2003). Also the EGF receptor antagonist Lrig1 has been proposed as a marker of human IF stem keratinocytes, since it maintains these cells in a quiescent state (Jensen and Watt, 2006). While these markers have greatly helped in isolating and characterizing keratinocyte subpopulations enriched in stem cells, a definite identification of epidermal stem cells and their niche is still vague. Clusters of putative stem cells expressing b 1 integrin, MCSP, and Lrig1 are mostly detected in the upper part of the epidermal rete ridge, whereas a bri 6 CD71 dim a keratinocytes are found at the tip of the rete ridges. On the other hand, xenograft studies using lentiviral vectors have shown that stem cells are dispersed throughout the basal layer Taichman, 2001, 2005). Irrespective of the markers and the location, stem cells can be identified based on the fact that they divide infrequently, thus maintaining the incorporated radioactive thymidine for long periods of time. This provides  Anti-apoptotic markers are almost exclusively expressed in KSCs, whereas pro-apoptotic molecules are mainly detected in more differentiated cells.
JOURNAL OF CELLULAR PHYSIOLOGY E P I D E R M A L S T E M C E L L S evidence that at least mouse epidermis contains slow cycling stem keratinocytes. On the other hand, radioactive labeling cannot be applied to humans in vivo. Yet, using a scaffold-based organotypic skin culture, it is now possible to mark keratinocytes with iododeoxyuridine and follow them for 8-10 weeks, when only <1% of basal cells have retained the label. Using whole mount preparations, Muffler and co-workers were able to demonstrate that LRCs are individually dispersed throughout the basal layer in a random distribution with no evidence of a pattern suggestive of EPU. As a similar pattern was also observed for the IFE of mouse skin (Braun et al., 2003), this is likely to be the kind of distribution of stem cells in the basal layer. Consistent with this finding, we have shown recently that survivin identifies a population of IF stem cells expressing highest levels of b 1 integrin, and these cells are dispersed randomly in the basal layer (Marconi et al., 2007). Actually, one could comment that such a random distribution throughout the entire basal layer better fulfills the tasks of IF stem cells that would thus be ready to reconstitute epidermis wherever it is needed. While stem cells should reside in protected locations, factors, adhesion molecules, and cells participating in the niche are evenly expressed throughout the epidermis and not just at the tip or in the upper part of the rete ridges (Muffler et al., 2008).

Hair Follicle Stem Cells
In contrast to continuous regeneration of the IF epidermis, the HF undergoes cycles of growth (anagen), apoptosis-mediated regression (catagen), and quiescence (telogen) (Alonso and Fuchs, 2006). The new hair shaft is regenerated from hairspecific SCs resident in the so-called bulge region of the HF epithelium. The bulge provides a unique differentiationrestricted environment for different types of adult SCs. In fact, the bulge area also contains melanocyte SCs and SCs with neural crest characteristics (Yu et al., 2010). Major differences exist between rodent and human HFs (Cotsarelis, 2006). For example, as opposed to the mouse HFSC niche, the human bulge is less morphologically distinct. Another difference is that HFs grow for years in humans and for weeks in rodents. Moreover, human and murine follicular epithelial SCs have a different biomarker expression.
Human bulge cells express CD200, K15, and K19 as the most reliable markers, while CD34 and K15 are preferentially expressed by mouse bulge cells. In vivo, the mouse CD34 þ keratinocytes are label-retaining cells, while in vitro they show enhanced colony-forming ability (Trempus et al., 2003). Bulge cells targeted by the K15 promoter have revealed a gene expression profile substantially overlapping that of CD34 þ cells, consistent with stem cell properties (Morris et al., 2004;Trempus et al., 2007). Using different biomarkers, Inoue and coworkers have recently described two stem cell subpopulations in human HF: CD200 þ CD34 À K15 þbright cells are located in the basal layer and form few but large-sized colonies, while CD200 þ CD34 À K15 þlow cells are mostly found in suprabasal location and form smaller colonies (Inoue et al., 2009). Two distinct CD34 þ bulge stem cell populations with stem cell properties have also been isolated in the mouse HF based on a 6 integrin expression and on the attachment to the basal lamina (Blanpain et al., 2004). Recently, Lgr5 þ [leucine-rich repeat containing G protein-coupled receptor 5] stem cells were identified in the mouse telogen bulge and in the lower outer root sheath of anagen HF (Jaks et al., 2008). Lgr5 þ population was strongly enriched for bulge markers such as CD34 (Snippert et al., 2010). Although Lgr5 þ cells rarely include LRCs and actively proliferate, these cells act as fully functional HFSCs in vitro and in vivo but do not contribute to the SG or IF epidermis (Jaks et al., 2008). Stem cells can also be found in areas of the mouse HF located above the bulge: the isthmus, the infundibulum, and the SG. In the upper isthmus, resides a subpopulation of multipotent and highly clonogenic keratinocytes, expressing Lrig1 and Plet-1 markers, recognized by MTS24 antibody (Nijhof et al., 2006). MTS24 þ cells are distinguished by low a 6 integrin levels and are negative for CD34, K15, and Sca-1. On the other hand, Sca-1, that is a marker of mouse IFE, also identifies a population of cells in the HF infundibulum that can regenerate epidermis but not HF (Jensen et al., 2008). More recently, Snippert et al. have identified a Lgr6 þ cell population located in central isthmus of the HF, directly above CD34 þ K15 þ bulge cells. Lgr6 þ cells do not retain DNA label and overlap only marginally with MTS24 þ Lrig1 þ cells but give rise to all lineages of the skin. Moreover, Lrg6 þ stem cell pool can renew sebaceous cells and seed the epidermis throughout life (Snippert et al., 2010). A discrete population of progenitor cells has also been identified in the SG that arises from the hair follicle. SG stem cells are marked by Blimp-1, and conditional ablation of Blimp-1 induces SG hyperplasia. Lineage tracing experiments confirm that SG stem cells can give rise to the entire gland, thus ensuring SG homeostasis (Horsley et al., 2006). Finally, another reservoir of mouse stem cells has been identified in the junctional zone between the IFE, SG, and bulge. Junctional zone cells are a pool of quiescent stem cells identified by Lrig1 (Jensen et al., 2009).

Stem Cells and Apoptosis
KSCs reside in a special microenvironment, the niche, that allows them to maintain their unique features and ''stemness.'' Namely, IF KSCs are located within the basal layer of epidermis and rest upon the BM that is rich in extracellular matrix (ECM) proteins and growth factors. As basal keratinocytes exit the niche, they move into suprabasal layers, and different microenvironmental stimuli influence their destiny. Under normal conditions, basal keratinocytes leave the basal niche to undergo terminal differentiation. This critical process is regulated by an increasing number of signals. Notch ligands expressed in basal keratinocytes bind to Notch receptors in suprabasal cells, thus promoting the classical change between keratins 5/14 and K1/10 that characterizes the commitment of basal keratinocytes to differentiation (Wang et al., 2008;Watt et al., 2008). While these and other molecular markers finely regulate the program of epidermal differentiation some of these signals are also involved in apoptosis (Lippens et al., 2005), suggesting that, under different regulatory inputs, basal cells exit the niche to undergo cell death. Conceivably, the niche concept in itself implies that KSCs, in order to maintain longevity and to ensure continuous tissue renewal, need to be protected from apoptosis. Indeed, transition from KSC to TA cells is associated with increased rate of apoptosis both in the murine (Tudor et al., 2007) and in the human system (Tiberio et al., 2002). Moreover, K15þ stem cells are more resistant than TA cells to apoptosis in a murine model of graft versus host disease (Zhan et al., 2007), while epithelial cells with stem-like properties are less prone to cell death than the reminder population (Harper et al., 2010). DNp63, a member of the p53 family, is critical for epidermal commitment of embryonic stem cells (Medawar et al., 2008) and identifies epidermal stem cells (Senoo et al., 2007). Interestingly, KSCs lacking DNp63 undergo apoptosis, while down-regulation of DNp63 is required for UVB-induced apoptosis (Liefer et al., 2000;Senoo et al., 2007). Moreover, up-regulation of microRNA-203 that is associated with keratinocyte differentiation not only inhibits the DNp63dependent clonogenicity of epidermal stem cells (Yi et al., 2008) but it also induces KSC apoptosis (Lena et al., 2008). It is likely that during epidermal homeostasis, DNp63 represses a number of genes, including p53, and protects KSC from cell death, which in turn contributes to maintain ''stemness.'' Additional insights into the relationship between KSC and apoptosis come from the study of the integrin pathway. Indeed, both a 6 b 4 and b 1 integrin are highly expressed in KSC and, upon binding to ligands of the BM component, promote proliferation and survival (Manohar et al., 2004;Piwko-Czuchra et al., 2009). Blocking b 1 integrin induces KSC anoikis, an apoptotic signaling cascade caused by cell detachment from the ECM and characterized by down-regulation of Bcl-2 (Tiberio et al., 2002) as well as by the early activation of caspase-8 (Marconi et al., 2004). This indicates that KSCs are protected from cell death because of the high levels of b 1 integrin. Thus, it would appear that low levels of b 1 integrin are responsible for KSC to exit the niche and undergo either differentiation or apoptosis. While in epidermis these two events are clearly different at the molecular level (Lippens et al., 2005), it is interesting to note that blocking b 1 integrin induces keratinocyte anoikis without affecting the markers of differentiation and vice versa (Mitra et al., 1997;Tiberio et al., 2002). The role of b 1 integrin within the KSC niche is even more complex, since the b 1 B isoform cooperates with unligated b 1 A to induce anoikis (Lotti et al., 2010). The best example of the close correlation between KSC and apoptosis derives from survivin, an antiapoptotic molecule that also regulates cell cycle (Mita et al., 2008). Survivin identifies KSC, whereas it is almost undetectable in TA cells. When keratinocytes undergo anoikis induced by blockade of b 1 integrin, survivin protein disappears (Marconi et al., 2007), and survivin overexpression protects keratinocytes from UVBinduced apoptosis (Dallaglio et al., 2009). Although the molecular switch determining differentiation or apoptosis of KSC remains to be defined, it should be taken into account that epidermis is a unique tissue under the constant physiological stress of UV irradiation that is one of the major causes of apoptosis. Interestingly, UVB irradiation induces keratinocyte apoptosis mainly in the proliferative basal layer (Qin et al., 2002). The balance between KSC resistance to UVB-induced apoptosis (Dallaglio et al., 2009) versus the higher susceptibility of the TA populations is likely to play a role in several skin disorders, including skin cancer where KSC carrying UVBinduced p53 mutation resist to apoptosis and expand because of the cell death of adjacent normal keratinocytes outside their boundary (Chao et al., 2008). Little is known about the correlation between stem cells and apoptosis in HF. During catagen, the lower two-thirds of the hair follicle rapidly regress, mainly by apoptosis of inner and outer root sheath keratinocytes in association with low levels of Bcl-2 and highest levels of the p75 neurotrophin receptor (Lindner et al., 1997), while bulge hair follicle stem cells escape apoptosis (Schneider et al., 2009). Recently, Tiede et al. (2009) have been able to select vital K15 þ progenitor keratinocytes from the adult hair follicle by a number of criteria, including resistance to apoptosis, and high levels of Bcl-2 mediate resistance of hair follicle bulge stem cells to DNA-damage-induced cell death . On the other hand, alopecia areata is caused by induction of apoptosis in bulge stem cells, and premature ageing in mice is characterized by increased bulge stem cell apoptosis (Espada et al., 2008).

Stem Cell, Apoptosis, and Skin Cancer
KSCs are long-lived residents in the epidermis, do not protect their genome by asymmetric chromosome segregation (Sotiropoulou et al., 2008), and are highly resistant to apoptosis. It is thus likely that they accumulate a number of oncogenic mutations and induce skin cancer formation. It has been shown recently that the same cell subset characterized by high clonogenic potential and by a great resistance to apoptosis is common to both normal and malignant epithelial cells (Harper et al., 2010). In human skin, survivin and its antiapoptotic isoforms are mostly expressed in KSC, whereas the proapoptotic variants are only expressed in more differentiated cells (Marconi et al., 2007). Survivin is abundantly expressed in squamous cell carcinoma (SCC; Lo Muzio et al., 2001) that is caused primarily by chronic exposure to UV irradiation and originates from KSC (Zhang et al., 2005). Upon sustained UVB irradiation, KSC, highly expressing survivin and carrying p53 mutations, colonize the adjacent epidermal compartments (Zhang et al., 2001). Moreover, skin specific expression of survivin promotes new clonal formation and malignant progression in an UVB-irradiated transgenic mouse model (Zhang et al., 2005). This suggests that survivin, highly expressed in KSC, contributes to skin carcinogenesis, by protecting cells from UVB-induced apoptosis (Dallaglio et al., 2009). Consistently, Bcl-xl plays a key role in UVB-induced skin carcinogenesis also through the up-regulation of survivin in cells of the bulge region (Kim et al., 2009a). Moreover, Stat3, a transcription factor that regulates the expression of several antiapoptotic genes, including survivin and Bcl-xl, is involved in a number of human tumors. Specific disruption of Stat3 in bulge region KSC inhibits skin tumor initiation by increasing carcinogen-induced apoptosis (Kim et al., 2009b). This indicates that Stat3 is required for the survival of bulge region KSC during the initiation of skin tumors. SCCs are tumors that arise in multilayered epithelia such as the epidermis, esophagus, cervix, and oral cavity. Independent of the tissue of origin, SCC is characterized by changes in integrin expression (Janes and Watt, 2004), and b 1 integrin mutation increases conversion of benign to malignant skin cancers (Ferreira et al., 2009). In the epidermis, SCC originates from KSCs (Klein et al., 2009) that are protected from anoikis because of the highest levels of b 1 integrin (Tiberio et al., 2002). While anoikis is important for tissue homeostasis, anchorage-independent growth is a crucial step during tumorigenesis and in particular during the metastatic spreading of cancer cells (Chiarugi and Giannoni, 2008). Conceivably, upon carcinogenic stimuli, integrin expression is perturbed and the BM is extensively damaged (Savoia et al., 1994;Ferreira et al., 2009). Under these conditions, KSCs detach from the BM, proliferate, and may reattach to an inadequate location to eventually metastasize. We have recently shown that the b 1 B-integrin isoform can cause apoptosis, possibly working as a ''sensor'' of the detachment from the proper ligand, thus acting as a ''fail-safe'' device by stimulating cell death to prevent spread of SCC (Lotti et al., 2010).

Conclusions
KSCs are distributed in different pools within adult epidermis, where they play a critical role because of their self-renewal, high proliferative, and multipotent differentiation capacity. These features have been already exploited for regenerative medicine and tissue repair (Pellegrini et al., 1999) as well as for gene therapy in genetic skin diseases (Titeux et al., 2010). Moreover, alterations of stem cells and their niche are of great relevance to the ageing process (Giangreco et al., 2008). On the other hand, despite a great deal of research, a number of questions still remain unanswered. First, no definite marker has been identified yet that allows the isolation of a pure population of KSC. Secondly, the role of KSC progeny, namely of TA cells should be better defined. Clayton and co-workers have recently challenged the existence of TA cells, by proposing a model whereby a single cell population has three stochastic choices upon cell division (Clayton et al., 2007;Jones et al., 2007). On the other hand, TA cells have always been overlooked and there are no clear markers to identify them. According to the early methods of enrichment based on the levels of b 1 integrin, cells not adhering to the ECM within 20 min are considered TA cells (Jones and Watt, 1993). Our group has modified this technique and has defined TA cells as cells adhering in a longer period of time, while cells not adhering were considered as postmitotic (Tiberio et al., 2002). Because of the lack of markers for TA cells, it is also difficult to assess their precise location within the basal or suprabasal layer. It is interesting to note that data are now emerging on the relevance of TA in physiological and pathological processes. It would appear that aging epidermis is characterized by changes in TA, but not in stem cell kinetics (Charruyer et al., 2009). In psoriasis, a skin disease characterized by altered epidermal homeostasis, TA, and not KSC seem to carry the intrinsic defect leading to the development of the lesion (Franssen et al., 2005;Grabe and Neuber, 2007). While the microenvironment of HF KS seems to be better characterized, the IF KSC niche needs to be fully understood mostly because KSCs are individually dispersed throughout the basal layer that in turn could be considered itself as a niche (Fuchs, 2009). This would imply that adjacent keratinocytes, other skin cells, adhesion molecules, and growth factors in any area of the basal cell compartment could contribute to the niche. In this respect, we have shown that a complex neurotrophin network exists in the skin and perform a number of autocrine and paracrine functions (Botchkarev et al., 2006). In particular, nerve growth factor is highly expressed only in KSC (Marconi et al., 2003) and stimulates KSC, but not TA cell proliferation (personal observation), suggesting an autocrine loop that could participate in the KSC niche. Finally, although many evidence point to KSC as the origin of skin cancer, more studies are needed to fully define both the stem pool involved in each type of tumor and the pathways through which a stem cell drives carcinogenesis, also in relation to UVB. Interestingly, a recent work points to the IF KSC as the origin of basal cell carcinoma (BCC; Youssef et al., 2010), as opposed to the dogma that the cell of origin of BCC resided in the HF bulge. This is actually conceivable, since BCC is a UV-induced tumor and IF KSC are closer than HF KSC to the skin surface.
In conclusion, a better knowledge of the different KSC pools in relation to their niche and to apoptosis will allow to design new therapeutic strategies either to protect ''stemness'' and allow tissue repair and regenerative processes or to target KSC when they acquire mutations and give raise to skin carcinogenesis.