Molecular Conversations and the Development of the Hair Follicle and Basal Cell Carcinoma Free

Over the last 30 years, numerous publications have described how human hair develops in both fetal and postnatal life. Over the last 10 years, lessons learned about hair development have led to investigations into the origin of human basal cell carcinoma (BCC) of the skin. Such investigations are shedding light on the central role of embryonic pathway signaling, particularly Sonic hedgehog (Shh) signaling and interactions with the insulin-like growth factor (IGF) axis, in this cancer.

In the embryo, the dermis provides a primary signal, causing epidermal cells to develop an epithelial bud, also called a placode and hair bud, which will develop into the outer root sheath of the hair follicle. The epithelial bud signals back to the dermis, which then causes mesenchymal cells to condense and form the dermal condensate or mesenchymal dermal papilla (1). The mesenchymal dermal condensate provides a secondary signal to the epithelial bud in the epidermis to form the hair follicle sheath and to descend with the mesenchymal dermal papillae into the dermis.

The fetal hair follicle has a region known as the “bulge,” which contains stem cells with the capacity to differentiate into various cells of the hair follicle, sebaceous gland and duct, apocrine gland and duct (in some hair follicles), and epidermis (2). It also has a region known as the “hair germ,” or “hair matrix,” which contains rapidly proliferating cells that develop the inner root sheath and hair shaft (1, 3).

Postnatal hair growth recapitulates much of embryonic hair growth (Fig. 1). Lying at the base of the outer root shaft of the follicle, “der Wulst,” or “the bulge,” is analogous to the embryonic bulge. It is also composed of pluripotent stem cells (capable of differentiating into different types of mature cells; ref. 3). Also lying at the base of the outer shaft but in an area distinct from the bulge, the germinative matrix is analogous to the embryonic hair germ (4). The base and lateral walls of the postnatal hair follicle envelop the dermis, and together they descend into the dermis and ultimately to the fat.

The close physical relationship between the mesenchymal dermal papillae and the developing hair follicle facilitates molecular signaling between these two regions. This “conversation” is critical for regulation of normal hair follicle maturation, both in utero and in postnatal life; it occurs primarily between proteins and genes in well-described embryonic pathways. The best-described pathways are Shh, Wingless-related mouse mammary tumor virus integration site (WNT; ref. 5), bone morphogenic protein (BMP), and Notch.

Shh signaling from the dermis is required for hair follicle development in the fetus. The critical role of Shh in this “organ” was first described in the late 1990s (6, 7). In postnatal life, immediately before and during anagen, the first and longest phase of the adult hair cycle, during which there is active growth of the follicle, the hedgehog ligand Sonic is produced by dermal cells around the base and lower lateral walls of the developing hair follicle. The other hedgehog ligands, Indian and Desert, are not produced. Sonic ligand binds to the patched homologue 1 (PTCH1) transmembrane receptor and, to a lesser extent, to the PTCH2 transmembrane receptor. Ligand binding inactivates the Ptch1 receptor, resulting in relocalization of Smoothened (SMO) from the cell surface to the tips of cilia; this leads to downstream signaling events, including the activation of the glioma-associated oncogene homologue 1 (GLI1) proteins (8). Activated GLI1 transcription factors translocate to the nucleus and promote transcription of several genes, including cyclin-dependent kinases D and E (cyclin D and cyclin E), vascular endothelial growth factor (VEGF), MYC (Myc), IGF binding protein 2 (IGFBP-2), and IGF-1 (Fig. 2). The hedgehog-interacting protein (HIP), a membrane glycoprotein, was recently shown to attenuate Shh signaling in a manner analogous to that of ligand-free PTCH1 (9). Shh signaling stimulates bulge stem cells to produce transit-amplifying cells, which undergo a limited number of divisions in the germinative matrix before differentiating (10). Without Shh signaling, the latter stages of hair follicle development would not occur. Shh signaling does not occur during catagen, the transition and shortest phase of the adult hair cycle, during which the hair club is formed, nor during the subsequent intermediate-length resting telogen phase, which results in the final keratinized, dead hair club.

The mesenchymal dermis ligands of the WNT family activate the canonical WNT pathway, causing an increase in nuclear β-catenin, which, along with other proteins, upregulates Shh signaling and Ptch in new follicles (11, 12). Overexpression of nuclear β-catenin seems to help determine the type of mature progeny that develop from progenitor cells, a process known as lineage choice. Generally speaking, signals from the WNT pathway activate the Shh pathway. There is evidence, however, that when the Shh pathway is ectopically activated, the reverse occurs, with Shh signals activating the WNT pathway (13). Noncanonical WNT signaling also occurs. There is a dramatic upregulation of WNT 10b at the earliest stages of embryonic hair follicle development and at the postnatal onset of anagen. WNT 5a seems to require Shh for its expression in developing hair follicles. WNT 10a and 10b are present in the hair follicle placode. WNT 3 is expressed in differentiating hair shaft precursors (14).

The Notch pathway is also important in hair follicle development in both the fetus and after birth and has been studied in this regard most extensively in inner-ear hair follicles (15). In the inner ear, especially in mesenchymal cells, Jagged 1 and 2 and Delta ligands from supporting cells bind to the Notch receptor, leading to γ-secretase cleavage of the receptor and release of the Notch intracellular domain into the cytoplasm. The Notch intracellular domain translocates to the nucleus, converting the centromere binding factor-1 repressor complex into an activating complex that upregulates the basic helix-loop-helix transcript regulators, hairy and enhancer of split (HES), and HES-related genes. HES and HES-related protein genes antagonize the mouse atonal homologue 1 gene (Math1, Atoh1; ref. 16).

BMPs are secreted molecules belonging to the transforming growth factor-β (TGF-β) superfamily. Various genes for BMPs and their receptors are found and seem to play a significant role in developing hair follicles in the mouse. During hair follicle induction, BMP-2 and BMP receptor-IA (BMPR-IA) are expressed in the hair placode; BMP-4 and noggin (whose gene product is an antagonist of BMP) are expressed in the mensenchymal condensate (17, 18). In postnatal life, BMP-4 is expressed in the dermis and germinative hair matrix; upregulation of noggin in the follicle and mesenchyme and downregulation of BMPR-IA in proliferating cells of the germinative matrix accompany the transition to anagen (19).

IGF-I and IGF-II are expressed by mesenchymal cells of the dermis and dermal papillae, and IGF-1 receptor (IGF-IR) gene expression has been identified in proliferating cells of the hair matrix (20, 21). The mRNAs of IGFBPs 3-5 are widely expressed in fetal mesenchymal skin cells; IGFBP-2 mRNA, on the other hand, is highly expressed in fetal epithelial cells (20). Together, these findings raise the possibility that IGF pathways, like embryonic pathways, or IGFBPs themselves may contribute to the regulation of hair follicle development.

Currently, it is generally believed that BCC develops from Shh-activated hair follicle progenitor cells that proliferate aggressively but fail to differentiate. It is also possible that Shh-activated epidermal stem cells, rather than hair follicle progenitor cells, are responsible for basal cell carcinogenesis (22). Patients with nevoid BCC syndrome have a defect in the Ptch1 gene located on chromosome 9q (9q22.3). They inherit a single defective copy of Ptch1 and subsequently suffer a second somatic hit, which results in loss of wild-type Ptch1 (23). Patients with sporadic BCC have several mutations in the Shh pathway, the most common being mutations in Ptch1. Deletion or inactivation of Ptch1 occurs in 30% to 60% of sporadic BCC (24). Activating mutations in SMO are well described and, although mutations in Gli1, Gli2, and Gli3 are relatively rare, overexpression of Gli1 is common (25–27).

New insights into the roles of IGFBP-2 in BCC are reported elsewhere in this issue of the journal (28). Keratin 14 (K14)-Cre: Ptch1^lox/lox mice, which have a deletion of Ptch1 in rapidly proliferating K14+ cells, are shown to have the following characteristics: ectopic activation of hedgehog signaling; early postnatal loss of hair follicle structure; expansion of the population of K14+ cells in the permanent region of the outer-root sheath around the hair follicle bulge; expansion of the population of K15+ bulge stem cells; increased numbers of P63+ transit-amplifying progenitor cells; expansion of the interfollicular epithelial basal cell layer; and, later in the postnatal period, the development of BCC. K14+ cells in this murine model have increased expression of IGFBP-2.

The expanded populations of K15+ and P63+ cells in the hair follicle, but not P63+ cells in the interfollicular epithelium, remain undifferentiated. Data support the hypothesis that Ptch1 deletion in K14+ cells increases the pluripotent stem cell niche and inhibits differentiation, whereas IGFBP-2 may increase epidermal progenitor cells and indirectly inhibit hair follicle proliferation (28). When IGFBP-2 is blocked, the epidermal progenitor-cell population fails to expand. These undifferentiated, proliferating cells are responsible for the formation of BCC, which also expresses IGFBP-2. As in K14-Cre: Ptch1^lox/lox mice, this report shows that IGFBP-2 was overexpressed in 20 of 23 human BCC samples. Ptch1 mutations in BCC are suggested by analogy to be associated with IGFBP-2 overexpression and BCC induction in humans (Fig. 2). The occurrence of IGF signaling in mice, however, could not be shown (29, 30).

IGFBP-2 is probably secreted into the extracellular space around K14+ cells, which is not discussed in the new report (28). Other researchers have shown the extensive interactions between Shh activation and downstream target genes in murine mesodermal-cell experiments; these downstream target genes regulate the Shh pathway and provide cross-talk between embryonic signaling pathways (29, 30). In the murine mesodermal cell experiments, Shh pathway signaling caused upregulation of the following genes: hairy and enhancer of split 1 (Hes1), which is critical in Notch signaling; HIP, which is important in the WNT and Shh pathways; sclerosin domain-containing 1 (Sostdc1), which is important in the WNT and BMP pathways; phosphatidic acid phosphatase type 2B (Ppap2b), important in the WNT pathway; and BMP-5 in the BMP pathway. Hedgehog activation causes downregulation of the following genes: BMP and activin membrane-bound inhibitor (Bambi), an inhibitor of the BMP pathway; secreted frizzle-related protein 1 and 2 (Sfrp2) in the WNT pathway; and WNT- 5b. One of the ways in which upregulation of Shh genes influences cross-talk with embryonic pathways is through the proteins they secrete. Therefore, it would be interesting to better understand whether, and how much, IGFBP-2 is secreted into the extracellular space around K14-Cre: Ptch1^lox/lox cells. This information could lead to a deeper understanding of how IGFBP-2 regulates embryonic pathways. Furthermore and of importance, demonstration of secreted IGFBP-2 around K14-Cre: Ptch1^lox/lox cells would potentially provide another avenue for understanding the relationship between IGFBP-2 and IGF pathway modulation in BCC. This issue deserves future exploration.

The consideration of why IGFBP-2 is localized where it is within the cell could also be informative. Does intercellular IGFBP-2 limit the availability of IGF-1, a putative mitogen (31)? Is there a receptor for IGFBP-2 on the BCC cell membrane, or is IGFBP-2 endocytosed? Furthermore, although the authors note IGFBP-2 expression in both mesenchymal and epithelial cells of K14-Cre: Ptch1^lox/lox mice, they do not elaborate on this expression, for example, by hypothesizing about the relative significance of IGFBP-2 expression in these two cell types.

It is well accepted that the genes for IGF and IGFBPs are upregulated by GLI transcription factors in the nucleus of Shh pathway–activated cells. The elegant experimental data appearing elsewhere in this issue of the journal (28) support an interesting link between the Shh pathway, Ptch1 expression, and Igfbp-2 expression in the evolution of BCC. We hope that future experiments will help provide answers to the important questions raised here.

No potential conflicts of interest were disclosed.

We thank Ronald Warren, PSI International, for editorial assistance and Christopher Risch, Technical Resources International, for assistance with the figures.