Keratinocyte Stem Cells and the Targets for Non-Melanoma Skin Cancer

Abstract

The mammalian skin is a complex dynamic organ composed of thin multi-layered epidermis and a thick underlying connective tissue layer dermis. The epidermis undergoes continuous renewal throughout life. The stems cells uniquely express particular surface markers utilized for their identification, isolation, and localization in specific niches in epidermis as well as hair follicles. The two stage skin carcinogenesis model involves stepwise accumulation of genetic alterations and ultimately leading to malignancy. Whereas, early research on skin carcinogenesis focused on the molecular nature of carcinogens and tumor promoters, more recent studies have focused on the identification of the target cells and tumor promoting cells for both chemical and physical carcinogens and promoters. Recent studies support the hypothesis that keratinocyte stem cells are the targets in skin carcinogenesis. In this review, we discuss briefly the localization of stem cells in the epidermis and hair follicles, and review the possibility that skin papillomas and carcinomas are derived from stem cells, as well as from other cells in the cutaneous epithelium whose stem cell properties are not well known.

INTRODUCTION

Benign and malignant cutaneous neoplasm can be induced on the backs of susceptible mice following a subthreshold exposure to a carcinogen (initiation) and subsequent chronic regenerative hyperplasia of sufficient magnitude (promotion). Initiation involves transformation of some of the epithelial cells into latent neoplastic cells whereas promotion elicits expression of the neoplastic change (1–4). Whereas early research on skin carcinogenesis focused on the molecular nature of carcinogens and tumor promoters, more recent studies have focused on the identification of the target cells for both chemical and physical carcinogens and promoters (5). Conceivably, any keratinocyte capable of proliferation could become and remain initiated. However, there is growing evidence that keratinocyte stem cells are the targets in skin carcinogenesis (6). Nevertheless, there is also recent evidence that terminally differentiated cells may retain an ability to form tumors. Here, we discuss briefly the localization of stem cells in the epidermis and hair follicles, and review the possibility that skin papillomas and carcinomas are derived from stem cells, as well as from other cells in the cutaneous epithelium whose stem cell properties are not known.

LOCALIZATION OF STEM CELLS IN EPIDERMIS AND HAIR FOLLICLES

The mouse epidermis is a continually renewing tissue, and especially the hair follicles undergo cyclic anagen, catagen and telogen phases. These dynamic changes make the location of epidermal stem cells complex and raise questions about the types of target cells for tumor initiation in skin. Even with the recent advancement in the field of skin stem cells and their localization, it is still controversial that which stem cell populations are the major target of carcinogen or tumor initiation [see Table I and Figure 1 for the details of different stem cell populations (7–8, 6, 9–13, 11, 13–18)].

Table I.

Stem cell markers, their properties and location in mouse skin

Name of marker/Stem cell population	Location in skin	Additional remark	Stem cell properties	References
Interfollicular epidermis (IFE) stem cells	Interfollicular epidermis	Marked by their retention of tritiated-thymidine nuclear label	Slowly cycling Label retaining cells(LRCs)	(7)
LRCs	Bulge region of hair follicle (HF)		Slow cycling, LRCs	(8)
Epidermal proliferative unit (EPU)	Basal cells of epidermis and overlaying suprabasal cells	Best observed with acetic acid and Carnoy’s fixative in epidermal sheets	Slowly cycling, LRCs are central cells, pulse labeled are periphery	(6)
EPU	Sebaceous gland (SG)		Long lived progenitor cells	(9)
Keratin-15 (K-15)	Bulge region of hair follicle of mice and human		Capable of generating epidermis, hair follicle and sebaceous gland	(10)
CD34	Bulge region of mouse, external root sheath in human	Expression restricted to bulge regardless of hair cycle stage	Used for enrichment of hair follicle bulge keratinocytes/able to reconstitute epidermis	(11, 12, 13)
CD34 and α6- integrin	Bulge region	Endothelial and mast cells also expressCD34/α6-integrin + cells are enriched for LRCs with high nuclear to cytoplasmic ratios	Specific for keratinocytes/slow cycling/colocalise with LRC/high proliferative potential in culture/make large colonies compared to CD34-cells	(11, 13)
Leucine-rich repeat-containing G protein (Lgr5+) cells	Bulb region	Lgr5+ marks stem cells in small intestine, colon and hair follicles	Contribute to all hair lineages but not to SG and IFE	(14)
Lgr6+ cells	Isthmus	These are not the LRCs	Prenatal Lgr6+ cells can form HF, SG, and IFE	(15)
Mts24	Region between sebaceous glands and hair follicle bulge	Does not express bulge specific CD34 and K-15 marker	Expression pattern similar to Cd34 cells/show colony phenotype	(16)
Blimp1	Sebaceous Gland	Blimp1 regulate size, cell number of sebocytes in SG, or role in SG homeostasis	Unipotent progenitor cells	(17)
Nestin	Bulge of vibrissa hair follicle	K15+and Nestin+, can form neurons after transplantation to the subcutis of the nude mice	Multi-potent stem cells in vitro	(18)
Stem cell population in matrix	Matrix		Controversial/debate

Figure 1.

The figure is showing the various parts of hair follicle in skin, and localization of the major stem cell population in hair follicle.

Abbreviation: (Bu = bulge region, SG = Sebaceous gland, APM = Erector Pilli muscle, DP = Dermal papillae, IFE = inter-follicular epidermis, IFD = Infundibulum, ORS =Outer root sheath, IRS = Inner root sheath, Lgr6+ = Leucine rich repeat containing G protein-coupled receptor 6, k-15 = keratin =15).

Gilbert and Lajtha (1965) proposed a hierarchical model for cellular replacement in the hematopoietic bone marrow where stem cells are renewing cells that also produce proliferating transit amplifying cells that undergo limited divisions prior to terminal differentiation (19). Under steady state conditions, stem cells are thought to be quiescent or to cycle slowly and to be protected within the tissue architecture. When stimulated, however, stem cells are capable of considerable proliferation. Additionally, stem cells posses many properties such as they are relatively undifferentiated (ultrastructurally and biochemically); they have high proliferative potential and are responsible for long-term tissue maintenance of the tissue; they may be slow cycling presumably to conserve the proliferative potential and to minimize the DNA replication error; they are often located in close proximity to rapidly dividing cells; and are usually located in well vascularized and well protected area (reviewed in 8). Furthermore, some but not all stem cells may be multipotential and capable of producing more than one differentiated lineage. Application of this concept to the cutaneous epithelium has been the subject of a number of reviews (16, 20–21).

The idea that the epithelium of the skin had stem cells grew from the observation of patterns of proliferation within morphological units of structure that in the epidermis are termed “epidermal proliferative units” (EPUs). These EPUs were first observed in the rodent ear by Mackenzie (1969) and others (7, 22–23), and were later named by Allen and Potten (1974) (24). Whereas the basal cells at the periphery of the EPU readily labeled with [3H] thymidine and were frequently characterized by mitotic figures (24–25). The central cells in the EPU required continuous labeling with [3H] thymidine in order to label and were termed label retaining cells (LRCs) because they persisted in the EPU for weeks to months following labeling (7, 26–28) These observations together with mathematical modeling studies (29), suggested that the central cell might have properties of stem cells. Indeed, Morris and Potten, (1994) demonstrated that [3H] thymidine label-retaining cells in vivo behaved in vitro like clonogenic stem cells whereas pulse labeled cells in vivo rarely formed colonies in vitro (6).

The LRC concept was extended the hair follicle by Cotsarelis et al. (1990), who found LRCs in the middle, permanent region termed the bulge (8). Subsequently, Oshima et al. (2001) demonstrated by micro dissection of hair follicles and grafting techniques that bulge cell were enriched for in vitro clonogenic cells and that bulge cells could differentiate into all cell types in the cutaneous epithelium (30). More recently, Morris et al. (2004) and Tumbar et al. (2004) and employed transgenic mouse technology to trace in intact cycling pelage follicles the lineage of bulge cells into differentiated epithelia, and demonstrated that bulge cells have a distinctive gene expression signature (10, 31). It is now known that hair follicles are home to multiple populations with stem cell properties, each with different markers and different locations within the follicle; see Table I, Figure 1).

Fluorescence activated cell sorting (FACS) in conjunction with a cell culture, grafting, or gene expression studies has also been an important technique for identifying keratinocyte stem cells in the cutaneous epithelium. To this end, Jones and Watt (1993) demonstrated that keratinocyte stem cells may be selected by their relatively high level of beta-1 integrin expression (32). Bickenbach and Chism (1998) also used beta-1 integrin expression to enrich for LRCs based on their greater adhesiveness to type IV collagen (33). Alpha-6 integrin expression together with reduced expression of Cd71 was used by Tani et al. (2000) to enrich for small and relatively undifferentiated keratinocytes that turned out to be LRCs (34). Then, Trempus et al. (2003) discovered that living cells of the hair follicle bulge expressed both alpha-6 integrin and the hematopoietic stem and progenitor marker Cd34 (13). This population was enriched for LRCs and not PLCs that produced large colonies in culture. Tumbar et al. (2004) used a strategy that enriched for hair follicle bulge cells based on their expression of histone H2b-Green Fluorescent Protein (31). Blanpain et al. (2004) also used Cd34 for FACS sorting, but found two populations based on the presence or absence of immunoreactivity with alpha-6 integrin (11). Although these sorting strategies effectively isolate living bulge cells, strategies for sorting interfollicular stem cells have not been forthcoming. The Lrig1 is another stem cell marker which characterized an adult human interfollicular epidermal stem cell and defines the hair follicle junctional zone adjacent to the sebaceous glands and infundibulum in mouse (35).

Additionally, in the bulge region of hair follicle, a population of CD34⁺ cells is distinguished by expression of Lgr5 marker, originally identified as a marker of intestinal stem cells. Lgr5⁺ cells can reconstitute all the epidermal lineages in skin reconstitution assays, in immunocompromised mice, and maintain the HF lineages under steady state conditions (14). However, lying just above the CD34⁺ bulge region are cells that express the MTS24 epitope. The MTS24⁺ cells are quiescent LRC, clonogenic, and express alpha-6 integrin but not K15 or CD34 (16) (See Table I). Yet another stem cell population between the infundibulum and bulge region of the hair follicle that expresses alpha-6 integrin, MTS24 but lacks CD34 and Sca-1 expression, and is able to reconstitute the IFE, HF and SGs (36) (See Table I). Similarly, Lrig1 is a marker of human IFE stem cells. Lrig1 cells are located in junctional zone between IFE, SG and bulge region in the mouse and express Lgr6 but not the Lgr5 and CD34, and are able to reconstitute the IFE and SC (35). Another distinct stem cell population exists in sebaceous gland that is Blimp1 (9, 17) (See Table I). Recently, a touch dome progenitor stem cell population (epidermal keratinocytes) distinguished by their unique expression of alpha-6 integrin, Sca1 and CD200 surface protein has been identified with the capability to make squamous and neuroendocrine epidermal lineages (37).

Finally, the lineage tracing strategies of Tumbar et al. (2004) and Morris et al. (2004) have been very productive towards visualizing the fate of cells from the hair follicle bulge. Both of these groups found that bulge cells contributed to the hair follicle cycling, but only rarely contribute to the homeostasis of the epidermis and sebaceous gland (10, 31). The Morris et al. (2004) study used a keratin 15 promoter to target mouse bulge cells with an inducible Cre recombinase construct or with the enhanced green fluorescent protein (EGFP) as a reporter. However, the study by Tumbar et al. (2004) used transgenic mice expressing H2B-green fluorescent protein controlled by tetracycline-responsive regulatory element (TRE) to trace the label retaining cells in hair follicle. This study also suggested that the bulge stem cell niche is a growth and differentiation restricted environment. Levy et al. (2003), using a different transgenic model, discovered that sonic hedgehog expressing follicular cells contributed to epidermis and sebaceous gland (38). Ito et al. (2005) expanded on these initial finding by studying wounding in the Krt1-15CrePR1;R26R mouse and found that, although the progeny of the keratin-15 expressing cells contributed to the early phase of wound healing, they did not persist in the healed wound (39). These findings are significant for the tumor studies in these mice described later.

CAIRNS HYPOTHESIS OF ASYMMETRIC STEM CELL DIVISION

Stem cell asymmetric cell division was proposed by Cairns in 1975 in which stem cell division produces one new stem cell and one transit amplifying (TA) cell (40). The hypothesis suggests that during every asymmetric cell division in a stem cell, the same DNA template strand will always co-segregate to form the DNA compliment of the new stem cell and the other strand (newly synthesized strand) remain with the TA cell. So with every cell replication one original copy of the DNA will remain the same in each subsequent cell division. In this condition, the daughter stem cell will always have same complementary strand of the DNA, which will be error free during the continuous process of cell division. In such condition, one can say that this ‘immortal’ DNA strand will pass down through stem cell generations. Additionally, the newly synthesized DNA strands will never remain in the stem cell compartment for more than one generation. In such a situation, the stem cell DNA strand may stay error free during DNA-strand replication errors. The daughter and granddaughter TA cells receive the newly synthesized DNA strand and may accumulate errors during DNA replication. However, the TA cells and their error containing DNA will sooner or later eliminated from the population. Thus, the mutations will not be accumulated in stem cells. The Cairns hypothesis of asymmetric cell division is supported in the crypts of the small intestine (41) and in an engineered mouse embryo fibroblast cell line (42). This idea of maintaining the integrity of error free DNA has profound implications in skin carcinogenesis.

IDENTIFICATION OF KERATINOCYTE SUBPOPULATIONS IMPLICATED IN SKIN CARCINOGENESIS

Although the concept that stem cells are the target cells (or cancer initiating cells) in skin cancer is not new (for example see 40, 43–44), there is now specific evidence that this is indeed the case in the cutaneous epithelium. A number of investigations have demonstrated a lifelong persistence of initiated cells in the epidermis and/or hair follicles (45, and references therein), which in light of the continual renewal of both epidermis and hair follicles, suggest that initiated cells may not be simply any proliferative cell, but keratinocytes with the persistence and self renewal properties of stem cells. Indeed, Morris and Potten (1999) identified a few cells in the middle, permanent region of the hair follicles that retain [3H] thymidine label for more than one year but remained in the follicle following plucking, and were even able to proliferate in response to it (46). Morris also observed radioactively labeled DMBA in the hair follicles for periods of up to three months (unpublished observations); however, later time points were not tested. In the case of the interfollicular epidermis, the central [3H] thymidine LRCs (label retaining cells) in the EPUs very clearly retained [14C] benzo[a]pyrene label for periods of about one month (47). It remains unknown to this day why the carcinogen label is retained. It is possible that the LRCs have as inherently low, or carcinogen induced, reduction in repair capacity relative to other basal cells, or whether some DNA adducts in effect hidden from the cellular repair mechanism, or whether some adducts are simply irreparable.

The second piece of evidence that LRCs might be players in epithelial cancer was that LRCs and not the more mature pulse labeled keratinocytes proliferated in response to the mouse skin tumor promoter 12-O-tetradecanoylphorbol-13-acetate (TPA) (27). Indeed LRCs or their daughters remained on the basal layer following their division to take up again a central position in the EPUs. In contrast, PLCs rarely divided, but instead underwent displacement from the basal layer and subsequent terminal differentiation. These results demonstrated that LRCs not only retained carcinogen-DNA adducts, but they also responded to TPA treatment by proliferating, a necessary event for tumor promotion.

Morris et al. (1991) then used density gradient sediment of freshly harvested keratinocytes to separate heavy and light subpopulations of basal cells, and demonstrated that the smallest and most dense subpopulation was enriched for in vitro clonogenic keratinocytes, for [3H] thymidine and [3H] benzo[a]pyrene label-retaining cells, for keratinocytes resistant to calcium-induced terminal differentiation from benzo[a]pyrene treated mice, for keratinocytes that proliferated in response to in vitro treatment with TPA, and for clonogenic keratinocytes (48). These results were important regarding skin carcinogenesis because they demonstrated that keratinocytes most relevant to skin carcinogenesis were different from the other, less dense basal cells not only in vivo, but also in vitro. Moreover, Baer-Dubowska et al. (1990) also observed that the cells from fraction 5 are enriched with B[a]P, DMBA, and (+)-anti-BPDE DNA adducts within a few hours after topical application (47, 49–50). This group also studied removal of adducts and found that keratinocytes from fraction 5 removed adduct more slowly than the more mature basal cells. These observations demonstrate that the labeled carcinogen was higher in these slowly cycling populations of epidermal cells ([3H] thymidine LRCs) in vitro and in vivo with stem cell properties. This observation suggests that epidermal stem cells may be at higher risk for induced initiation of cancer probably due to their properties of slow cycling and their ability to accumulate and retain DNA damage. Moreover, the epidermal cells are dynamic in nature and the differences in metabolic capabilities are observed among different epidermal subpopulations (51–54).

Also using density gradients, Battalora, et al. (2001) reported that the tumor promotion is age dependent in the Tg.AC (v-Ha-ras) transgenic mouse (55), and tumor promotion is correlated to changes in cellular development in the follicular compartment of the skin. The Ha-ras transgene in Tg.AC mice is highly expressed in dense basal keratinocytes. The transgenic expression was detected 9 days after 12-O-tetradecanoylphorbol-13-acetate (TPA) treatment, especially in most dense fraction of gradient-isolated keratinocytes. This was the earliest detection of transgene expression after the beginning of treatment. These results provide further evidence that the target cells of epidermal skin carcinogenesis are smaller and more dense than lighter basal cells (50). Additionally, a strong correlation was observed between a highly expressed cell surface marker β1-integrin [(a cell surface marker on epidermal stem cells) (56)] and Ha-ras transgene expression. It has been suggested that the early expression of the v-Ha-ras transgene could be located in hair follicle cells (57–58). Similarly, the hair follicle stem cell population is a slow cycling with relatively undifferentiated ultrastructurally, but can be initiated to proliferate transiently following treatment of skin with TPA (8). Such evidence provides clues that hair follicle stem cells may be involved in skin carcinogenesis.

The fourth piece of evidence that carcinogen target cells had characteristics of stem cells was the development of a sensitive and quantitative assay for clonogenic keratinocytes (45). This assay was placed in the context of delay interval skin carcinogenesis experiments designed to confirm the permanence of skin tumor initiation. Hence, the number of colonies remained essentially constant for more than one year following in vivo carcinogen exposure and well within the range of control-treated and untreated groups. This finding was important because it argued against a silent clonal expansion of stem cells following initiation in the absence of promotion. In contrast, following treatment of the control and initiated mice with TPA, the number of colonies increased with the number of TPA treatments in both groups; however, the number of colonies was always greater when the keratinocytes were harvested from initiated and promoted mice.

Morris et al. (1997) further investigated the stem cell properties of initiated keratinocytes by treating mice either before or after carcinogen exposure with 5-fluorouracil, an agent known to kill cycling but not quiescent cells (59). 5-fluorouracil applied topically twice to mice resulted in profound epidermal atrophy with subsequent sloughing of the epithelium. However, despite this severe damage, both papilloma and carcinoma responses in the 5-fluorouracil treated mice were surprisingly similar to the vehicle controls. This finding suggested that the initiated cells were quiescent rather than actively cycling (59).

Because portions of the hair follicles appeared to survive the 5-fluorouracil treatment, we wanted to investigate further the origin of the cutaneous neoplasms. Most previous evidence has placed the tumor origin from the hair follicles (60–61); however, mathematical models suggest that there is at least one population of stem cells in the interfollicular epidermis (29). Morris et al. (2000) used the abrasion technique of Argyris, but applied the abrasion one time only following carcinogen exposure, and then allowed the epidermis to regenerate from the underlying hair follicles. Upon tumor promotion with TPA, the abraded mice developed papillomas and carcinomas (62). However, the number of papillomas in the abraded group was approximately half that of the unabraded controls while the number of carcinomas was similar for both groups. These results were consistent with the hypothesis that the target cells of carcinogen action were found principally in the hair follicles, but were also present in the interfollicular epidermis or follicular infundibulum.

Yet another tantalizing line of research was explored by Trempus et al. (2007), who found that the living follicular bulge marker, Cd34, was necessary for skin tumor development. Mice in which Cd34 had been knocked out failed to develop papillomas during tumor promotion, apparently because they failed to develop follicular hyperplasia in response to TPA (63).

The foregoing studies were predicated on the concept that keratinocyte stem cells were slowly cycling or quiescent. However, recent emerging evidence suggests that there are at least two kinds of stem cell populations 1) quiescent (i.e., out of cycle) and cells in low metabolic state, and 2) active i.e., in cell cycle stage and not able to retain the DNA label. Both of these stem cell populations co-exist in several tissues including hair follicle, intestine and bone marrow and are believed to be cooperative in their functional role (reviewed in 64). For example, in the hair follicle, CD34⁺Lgr5⁻ (slow cycling) and CD34⁺Lgr5⁺(actively cycling) stem cells, respectively, are located inside and next to bulge area (14), consistent with Lgr5+ stem cells (i.e., actively cycling stem cells) located adjacent to LRCs (i.e., slowly cycling stem cell populations) (65). Similarly, there are some stem cells that are long lived yet constantly cycling in intestine (66). On the basis of above observations, the hair follicle appears to contain both quiescent and active stem cell populations in separate yet adjacent locations.

CONCEPT OF CANCER STEM CELLS

The concept of cancer stem cells (CSC) has generated as much excitement and hope for cancer treatment as perhaps any other area of cancer research over last decade (67–69). Cancer stem cells are thought to self-renew and produce progenitor cells/daughter cells that can generate the all types of cells comprising the tumor. CSCs are probably very rare within the tumor (reviewed in 70), and resistant to chemotherapy, radiation, and even targeted molecular therapies via their relatively high expression of the multi-drug resistance genes, a known characteristics of many stem cells (reviewed in 71). Like adult stem cells, the cancer stem cells have properties of self renewal, niche preparation, epithelial to mesenchymal transition (EMT) and resistance to apoptosis (reviewed in 72). Additionally, the CSCs may cycle through the quiescence and active cell division depending on the change in tumor microenvironment, and creation of new niches during the metastatic process. Cancer stem cells have recently been demonstrated in non-melanoma skin cancer such as SCC in mice, and controlled by TGF-β and integrin/focal adhesion kinase (FAK) signaling (67). The TGF-β pathway is crucially important in human cancer as it act as tumor suppressor initially but promotes metastasis in later stages (73).

The niche or the microenvironment of stem cells plays an important role for maintenance of the stemness. A role for Vascular Endothelial Growth Factor (VEGF) and its receptors in the process of angiogenesis has long been known. However, more recent data demonstrated that VEGF creates a perivascular niche for CSCs in skin and affects CSCs through a neuropilin (Nrp1) autocrine loop. Nrp1 is a VEGF co-receptor expressed in cutaneous CSCs, and able to block the VEGF activity to promote stemness and renewal. Hence, VEGF-Nrp1 loop regulates the initiation and stemness of skin tumor (i.e., the ability to self renew and differentiate) (74). Additionally, the deletion of Nrp1 in normal epidermis prevents skin tumor initiation. This report explains the role of a niche component in deciding the fate of various cancer stem cells in tumor initiation. A similar perivascular niche is also described for brain tumor stem cells elsewhere (75). However, in the future, it will be interesting to see the effect of various growth factors and their role in maintenance and proliferation of CSCs.

HAIR FOLLICLES AND THEIR CONTRIBUTION TOWARDS SQUAMOUS PAPILLOMA DEVELOPMENT IN MOUSE: AN EXAMPLE OF KERATIN 15 (K15) EXPRESSING STEM CELL POPULATIONS

The precise identification of cancer initiating cells and the molecular events associated with the tumor initiation in epidermal papilloma and squamous cell carcinoma are unknown. However, it is widely believed that a major target cell of carcinogens resides in the stem cell compartment in cutaneous epithelium (69, 76–77). One of the most convincing arguments for the stem cell being the target cell is the permanent nature of the neoplastic lesion: that tumors can be elicited even with long interval between the process of initiation and promotion (45). Bulge keratin-15 expressing cells are proposed to be a site of origin for skin cancer due to their quiescent nature favoring the accumulation of mutational load in their genome (8). However, due to availability of different keratin promoters, inducible Cre mice, and transgenic technology, the epidermal lineage tracing is possible following Cre activation in different compartments of the pilosebaceous unit (78). The K15 promoter preferentially targets stem cells in the bulge and secondary germs (10, 79) and this marker can be used to target bulge cells in mouse skin cancer model to determine the contribution of these cells towards development of skin cancer. There are contrasting results in this regard.

Recently, we have observed that the contribution of K15 containing bulge stem cell populations of hair follicle towards squamous papilloma development in the Krt1-15CrePR1; R26R bigenic mouse (Rosa 26 reporter mice). Beta-galactosidase activity was determined in skin and tumor section from untreated mice and DMBA, TPA treated mice. We made following conclusions based on several observations. First, K15 expressing cells contribute to most of the papillomas by 20 weeks of TPA promotion. Second, the K15 progeny persist in papillomas and some malignancies for many months following transient induction of the reporter gene. Third, papillomas have heterogeneity in cellular composition as well as in their expression of the codon 61 signature Ha-ras mutations. Together, these finding indicate that the K15 expressing hair follicles cells contribute to cutaneous papillomas with long term persistence and a subset of which express the Ha-ras signature mutation characteristic of initiated cells (see Figure 2; Morris, unpublished data). These observations suggest a strong evidence for the role of stem cells in skin papillomas. However, the contribution of these cells cannot be ignored in other non-melanoma skin cancers such as basal cell carcinoma and squamous cell carcinoma.

Figure 2.

Picture showing the transgenic krt1-15CrePR1; R26R bigenic mice skin initiated with DMBA followed by induction with RU486 and promotion with TPA in skin section (left side). Right side of the figure is showing progeny of hair follicle stem cells in skin tumor (blue cells) in whole mount (A & C) and skin section (B & D). The white and black arrow showing presence of blue cells or progeny of k15 expressing stem cells in hair follicle.

BASAL CELL CARCINOMA AND ITS HAIR FOLLICLE CONNECTION

Basal cell carcinoma (BCC) is the most common human cancers, are so named due to their resemblance to basal keratinocytes and hair follicle cells in skin. Bulge K15- expressing cells are proposed to be a site of origin for NMSC (8), and their expression profiling data is available to make conclusion towards stemness and various gene signature/pathways in skin non-melanoma stem cells (10). In skin, the Shh pathway is important for maintaining the epidermal stem cell population, and development of sebaceous gland and hair follicle (Reviewed in 80). However, abnormalities in Shh signaling pathway molecules including Shh, Ptch1, Smp, Gli1 and Gli2 are the key contributors in development of BCCs. Interestingly, the expression of Shh, and its receptors PTCH1, Smo, and Gli-1 is also detected in human fetal epidermis as well as freshly sorted human putative epidermal stem cells (Reviewed in 80). The over- expression of Sonic Hedgehog, and mutations in tumor suppressor gene PATCHED (PTC) in human patients is associated with BCC (81).

Regarding origin of BCC in skin, there are contradicting results. One study suggests that the BCC did not arise from bulge cells of hair follicle (78), whereas another recent study has shown strong evidence that the BCC arise from hair follicle stem cells in irradiated Ptch1^+/− mice (82). In the Ptch1^+/− model develop BCC following treatment with UV or ionizing radiation (82). Interestingly, the Ptch1 (a receptor for hedgehog protein) and Smoothened or Smo (SMO) a hedgehog signaling pathway gene are mutated in BCC. Additionally, Ptch^+/− mice with p53 deletion accelerates the BCC formation, and K-14 expression becomes limited to keratinocytes (82).

In general, mouse models in which the hedgehog pathway is activated downstream of Ptch1, such as transgenic expression of SmoM2 and Gli2, produce hedgehog driven tumors predominantly in the tail and ears. However, Wang, et al. (2011) did not find any BCC at these sites as they also did not observe Smo expression in hair follicles of tail and ear in Ptch1^+/− mice. Such data support the view that the Smo protein expression is necessary for BCC tumorigenesis (82). This collective evidence suggests that the K15 expressing cells are the predominant cells of origin of Ptch1^+/− murine BCC, irrespective of p53 mutational status. The loss of p53 significantly increases the tumor initiating capacity of follicular cells and non-K15 expressing cells such as interfollicular epidermis (82). Transgenic mice such as Ptch1^+/−K15-Cre-ER2 YFp IsI/IsI (Y15 mice) or Ptch1^+/−K14-Cre-ER2 YFp IsI/IsI (Y14 mice), and for p53 deletion Ptch1^+/−K15-Cre-PR1 p53^fl/fl (PF15 mice) or Ptch1^+/−K14-Cre-PR1 p53^fl/fl (PF14 mice) were used in this study (82). However, K14-CREER/RosaSmoM2, shh-CREER/RosaSmoM2, K15-CREER/RosaSmoM2, and K19-CREER/RosaSmoM2 transgenic mice were used in the other study (78).

A recent report establishes a link between wounded epidermis and cancer connection in mouse. The mouse model used in this study was a conditionally expressing GLI1 and another mouse model with homozygous ptch1 activation (a condition mimicking human BCC condition). The wounding environment enhances epidermal tumorigenesis (BCC like lesion) by recruitment of hair follicle keratinocytes (83). Additionally, the full thickness wounding is necessary to induce keratinocyte migration from the bulge and the non-permanent part of the hair follicle to the interfollicular epidermis (83). Such studies will advance the field of stem cells and cancer, and in the future, will be able to trace the precise lineage of cancer cells. The story of BCC in terms of its origin will be clearer, once such studies will be clarified upon reproduction in other mouse models. Moreover, to understand the contribution of stem cells towards cancer progression and development it is also necessary to discover more new stem cell markers and to characterize them for their attributes and proper localization in the epidermal compartment.

SQUAMOUS CELL CARCINOMA AND ITS HAIR FOLLICLE CONNECTION

Squamous cell carcinoma is the second most common types of skin cancer after BCC. Several recent reports demonstrate that hair follicle stem cells are involved in squamous cell carcinoma (SCC). One report suggests that the SCCs probably arise from bulge stem cell niche of hair follicle but not from the Shh-expressing transit amplifying (TA) cells of hair follicle (84). However, physiological expression of oncogenic Kras can drive formation of hyperplasia and EMT; however, it seems that a second hit is required to develop SCC (84–85). It is also noteworthy that the bulge stem cells and other lineages are able to initiate papilloma formation in skin, but multiple genetic hits are required in context of oncogenic Kras for the development of invasive SCC (86). In addition to bulge stem cells, cells of interfollicular epidermis are also capable to giving rise to benign squamous tumors upon physiological expression of oncogene Kras^G12D without any exogenous stimuli such as inflammation, chemical abrasion or wounding (86). In this study the transgenic mice were conditionally expressed Kras mutant (G12D) and an inducible Cre recombinase (k14CREER/kras^LSL-G12D, k15CREER/kras^LSL-G12D, and K14CREER/kras^LSL-G12D/p53^fl/fl) in different epidermal lineages (for example, k14-CREER in all basal cells, K15-CREPR, K15-CREER, K19-CREER in bulge stem cells).

Another report provided evidence that the tumor initiating stem cells of SCCs are controlled by TGF-β and integrin/focal adhesion kinase signaling (67). In addition, the possibility of a pool of multiple cancer stem cells cannot be ignored in cutaneous SCCs as evident from two cancer initiating stem cell populations identified based on integrin, Cd34 and β1 markers (67). In addition, there are mutations reported in Sonic Hedgehog (SHH) signaling including PTCH (in Basal-cell carcinoma) (87–88), RAS (Squamous cell carcinoma) (89), TP53 (SCC and BCC) (90), and PTEN (Trichilemmoma, papilloma, SCC) (91) gene of mice and human are associated with various kind of tumors.

Generally, for cancer initiation and progression, multiple genetic alterations are required. Most of these genetic alterations/mutations are reported in proto-oncogenes, DNA repair genes and tumor suppressor genes. Mutations in these genes are associated with various tumors. The tumor suppressor genes are the negative regulators for tumors. One such negative regulator is p53. p53 is involved in cell cycle arrest and activation of apoptosis (92–93). Mutations in the highly conserved p53 gene have been detected in 50% of human cancers and in almost all skin cancer (94). Inactivation of p53 is important during induction of skin cancer by UV radiation. Approximately 71% of the p53 mutations are detected in aggressive and non-aggressive BCCs and SCCs along with the UV signature (95). Moreover, mutations and the level of p53 are well connected to non melanoma skin cancers (NMSCs) and associated with UV exposure, DNA damage, DNA damaging agents, pre-cancerous skin lesions, SCC, and BCC (reviewed in 96). The mutation spectrum of p53 is not only involved in NMSC but also plays an important role in melanoma and many other cancers.

Aside from lineage tracing (understanding the cell fate involved during cancer initiation and progression), and the role of p53 mutation, there is also a need to understand and identify the genetic and environmental factors/stimuli, pathways related to the stem cell proliferation, maintenance and regulation.

GENETIC APPROACHES FOR IDENTIFICATION OF KERATINOCYTE STEM CELL REGULATORY GENES

Because keratinocyte stem cells now have an undisputed role is the biology of the skin as well as wound healing and development of epithelial tumors, it is important to understand how they are regulated. To address this issue, we have used the power of mouse forward genetics and have used the number and size of in vitro keratinocyte colonies as a phenotype, and have shown that these phenotypes are genetically defined quantitative complex traits amenable to genetic dissection. Our genome-wide linkage analyses of C57BL/6 and BALB/c mice revealed novel quantitative trait loci (QTLs), Ksc1 and Ksc2, with highly significant association with colony size and number (97), which were shown to be genetically defined independent multigenic traits (98–99). Further fine mapping within the Ksc1 locus was associated with small keratinocyte colonies on mouse chromosome 9. We identified bone morphogenetic protein 5 (Bmp5) as a candidate stem cell regulatory gene, which is expressed predominantly in the bulge area of hair follicle (97, 100). Notably, Bmp5 deficiency increased chemically induced skin tumor incidence in the two-stage mouse skin carcinogenesis model (100). These data suggest that Bmp5 may regulate stem cell homeostasis and function as a tumor suppressor gene.

In addition, Bmps and their signaling have also been implicated in many types of human cancers (reviewed in 101). BMPR1A mutations were found in patients with juvenile polyposis (102). BMP6 was shown to be absent in benign prostate tumors, but strongly evident in primary tumors with secondary bone metastasis, implicating that BMP signaling also plays a role in migration and invasion of prostate cancer cells (103). These results validate our finding and strongly suggest that Bmp5 may represent one of the factors in the pathway regulating stem cell homeostasis. Further analyses of genes or gene clusters that regulate the proliferation and stemness of KSCs should improve our understanding of molecular mechanisms and pathways that regulate stem cell behavior and may yield new insights into the development of stem cell-based cancer therapeutic agents.

Moreover, identification of genes or gene clusters controlling the proliferation of keratinocyte stem cells and their stemness are necessary for understanding NMSC. Because it is hypothesized that the identification of factors that regulate stem cell cycling and quiescence may yield new insights into methods to render cancer stem cells susceptible to chemotherapeutic agents, or to block the cancer stem cell cycling permanently due to drug (68). This area is very new and critically important since skin stem cells lie at the root of the epidermis and the hair follicle and plays an important role in skin homeostasis. These stem cells are the targets of non-melanoma skin cancer and other hyper-proliferative skin diseases (104). The identification of the gene or gene cluster that regulates the formation of large keratinocyte stem cell colonies has the potential to change current treatments or preventative interventions in dermatology. If we knew what gene or genes were driving keratinocyte stem cell hyperproliferation or decline, then we could design targeted therapeutic agents for these disorders with fewer side effects than those currently in use. Moreover, identifying a specific keratinocyte stem cell regulatory gene and its pathway members, would lead to novel strategies to prevent skin damage and cancer susceptibility induced by exposure to ultraviolet light, and possibly new diagnostic tools for impaired stem cell function.

In addition to classical forward genetics, reverse genetics have been used successfully, both by identification of an important gene, such as beta-catenin, and knocking it down or over-expressing it and examining the phenotype, which in this case, was linked to tumor development (73). Additionally, several gene expression studies have proven insightful (105). Dang et al. (2006) examined the profiles of different stages during the promotion of skin tumors in mice undergoing two-stage carcinogenesis and identified many differentially expressed genes (105). Additionally, Schober and Fuchs (2011) employed this technology and found a molecular signature for cancer stem cells in SCC that included carcinoma signalling genes such as Tgfb1, Ptk2 (FAK), Itgb1 (β1), Kras, Vegfa, Tgfa, Igf2r, Braca1, Braca2, mdm2, Ereg, Lasp1 etc. (67).

ROLE OF NON-STEM CELL POPULATIONS DURING CARCINOGENESIS AND EPIDERMAL MAINTENANCE

The role of stem cells in cancer progression and cancer initiation is discussed above, but there are some studies where the roles of non-stem cell populations are discussed during the tumorigenesis and homeostasis in skin. The hallmark of cancer is uncontrolled proliferation of dividing cells harboring oncogenic mutations. However, differentiated, non-dividing epidermal cells with activated MAPK kinase 1 (MEK1) can initiate benign tumor formation (106). The tumor initiation in this process is driven by recruitment of undifferentiated cells to the proliferative compartment of the tumor via inflammation. These observations indicate that the role of environmental factors such as wounding and inflammation cannot be ignored during the process of tumorigenesis in multi-layered epithelium. Moreover, in presence of oncoproteins and co operation of Ha-ras and Bcl-2 during multistep carcinogenesis malignant transformation increases significantly (107). Similarly, TGF alpha and v-fos cooperation in transgenic mouse epidermis induces aberrant keratinocyte differentiation and stable papillomas (108).

Furthermore, several studies have demonstrated that all of the stem cell populations are able to make all lineages of hair follicle (see Table 1). Nevertheless, involucrin⁺ differentiated keratinocytes are also able to form multilineage skin epithelia. This study also suggests that the commitment to differentiation does not prohibit cells from re-entering the cell cycle, de-differentiation, and acquiring stem cell like properties (109). This study is a good example showing the existence of epidermal strategy, where a differentiated cell is able to maintain the homeostasis and regeneration.

CONCLUDING REMARKS AND FUTURE DIRECTIONS

The perspective presented here has shown that the non melanoma stem cell populations are the important cell types in the hair follicle. The skin stem cells are organized in a well defined niche in a different sub-cellular compartment of the mammalian skin. The role of skin stem cells in carcinogenesis and the response of carcinogens to stem cells still require further experimental validation in multiple sets of mouse strains in order to confirm the basic understanding of cancer target cells. Recent experiments have shed some light on the stem cell origin of some non-melanoma skin cancers such as SCC, papilloma and BCC. Many studies have been performed in mice of different genetic backgrounds to understand better the role of cancer initiating stem cell population in skin. However, it is also important to consider selection of mouse strains for carcinogenesis as the different mice strains respond differently to classical two stage carcinogenesis (2). Ultimately, the genetic history of various mouse strains is very complex in term of breeding, genetic background, targeted mutagenesis, and various mutations (110).

We have discussed several cases where the stem cells appear to be the target of carcinogens in non-melanoma skin cancer. However, these targets may vary depending upon the chemical nature of the carcinogen or drug molecule as is evident by the use of Doxorubicin, a widely used chemotherapy drug. There is a report showing the effects of Doxorubicin on the disruption of the hair follicle associated blood vessels network but not on the stem cells (111). This finding might also explain the reason behind the re-growth of the hair even after chemotherapy. It is important to note that all chemotherapy agents used in experimental as well as clinical applications have a broad spectrum of chemical structures and activities, so their intended effects will vary in skin, especially due to their action upon the stem cells.

During past two decades, progress has been in our understanding of the role of epidermal stem cells and their contribution towards cancer origin and progression. However, a number of questions remain to be addressed especially when making any conclusion regarding keratinocyte stem cells and their involvement in cancer initiation and progression. First, the factors driving hair follicle stem cells to respond to a particular carcinogen, wounding, or regeneration in mammalian skin are not clear. Second, what are the autonomous central regulators/genes/molecular pathways of hair follicle cycling, stem cell mobilization, non-melanoma stem cell maintenance, and carcinogen responsiveness in skin? Is there any signaling to or from distantly residing stem cells such as in bone marrow or other tissues? How important are specific cell surface markers or receptors in terms of their cancer causing or promoting potential in skin? Does a relationship exist between stem cell surface markers from one type stem cell to another type of stem cell in same organ or different one? Are stem cells niches shared for skin cancers or do they involve contribution from multiple cell types?

These questions illustrate the potentially complex role of skin stem cells and their involvement towards skin carcinogenesis complicated. However, newer technological advances will assist in resolving epithelial stem cells in NMSC. We believe that answers of the above questions, and new discoveries in the field will make progress towards cure for skin cancers.

Biographies

Ashok Singh was born in Pauri Garhwal hills of Uttarakhand, India. He passed his M.Sc. in Zoology from Allahabad University and Ph.D. in 2008 from Dr. S.K. Rath’s Laboratory, CDRI, Lucknow, where he was registered from Jawaharlal Nehru University, New Delhi, India. He is a member of Indian Genome Variation Consortium. In 2008, he was a research fellow in the department of Molecular Pharmacology and Experimental Therapeutics, Mayo Clinic College of Medicine, Scottsdale, AZ. Currently he is working as Hormel Fellow in Stem Cells and Cancer Laboratory, Hormel Institute, Univ. Minnesota, USA. His research interest includes keratinocyte stem cell biology and cancer biology.

Heuijoon Park received his Ph.D. in 2011 in Pathobiology and Cell biology program from Columbia University. He trained from Morris laboratory as a Ph.D. student and studied roles of skin stem cells and bone marrow stem cells during skin tumor development. Dr. Park is currently member of Tarakhovsky laboratory at Rockefeller University in New York.

Thaned Kangsamaksin received his Ph.D. in 2011 in Cell Biology and Pathobiology from Columbia University. From 2006 to 2008, he worked as a graduate research assistant in Rebecca Morris’ laboratory in the Department of Dermatology at Columbia University, focusing on the role of Bmp5 in keratinocyte stem cell regulation. His doctoral work in Jan Kitajewski’s lab involved Notch signaling in tumor angiogenesis. He is currently a postdoctoral fellow in the Kitajewski laboratory and continues to work on the Notch pathway and its targets in tumor angiogenesis and metastasis.

Anupama Singh received her B.Sc. in Zoology, Botany and Chemistry; M.Sc. degree in Zoology, and Ph.D. in Habitat ecology from Laboratory of Fresh water Biology, in Dr. O. P. Gusain’s lab, Department of Zoology and Biotechnology, HNB Garhwal University, Srinagar, India. Currently she is working as Hormel Fellow in Stem Cells and Cancer Laboratory at the Hormel Institute, University of Minnesota, USA. Currently, she is interested in keratinocyte stem cell biology.

Nyssa A Readio received her B.S. in Biology with minors in Chemistry and Psychology in 2008 from the University of St Thomas. She interned as an undergraduate researcher in the laboratory of Dr. Glenn K. Sherer where she gained experience in microscopic anatomy and developmental biology. In 2008 she accepted a Junior Scientist position with Dr. Rebecca J. Morris in the Laboratory of Stem Cells and Cancer at the Hormel Institute, University of Minnesota. Her research interests include stem cell biology, cancer research, genetics, tissue regeneration, intercellular interactions and signaling mechanisms, bone marrow cells and keratinocytes. She desires to continue on with research and to pursue higher educational opportunities.

Rebecca J. Morris received her Ph.D. in 1981 in Biology from Syracuse University. After working as a Research Associate with Thomas Slaga’s group at M.D. Andersons’ Science Park, she accepted a position at the Lankenau Institute for Medical Research in Wynnewood, PA. In 2001, Dr. Morris moved her program to the Departments of Dermatology and Pathology at Columbia University Medical Center in Manhattan, New York. She is currently Leader of the Laboratory of Stem Cells and Cancer at the Hormel Institute/University of Minnesota directed by Dr. Zigang Dong. Dr. Morris has maintained an interest in keratinocyte stem cells and cancer from her graduate work, and throughout this time, has been funded by grants from the ACS, NIAMS, and NCI.