Patched1 Inhibits Epidermal Progenitor Cell Expansion and Basal Cell Carcinoma Formation by Limiting Igfbp2 Activity

Basal cell carcinoma (BCC) of the skin is the most common form of cancer, with the majority being caused by mutations in the Patched1 (Ptch1) gene, leading to activation of the Hedgehog (Hh) signaling pathway. Hh signaling is implicated in many tumor types; thus, defining the mechanisms by which Ptch1 regulates tissue proliferation is of paramount importance. Here, we show that the key role of Ptch1 in the skin is to limit the size of the epidermal stem/progenitor compartment and allow hair follicle differentiation. Specifically, loss of Ptch1 leads to the promotion of progenitor cell fate by increasing basal cell proliferation and limiting the progression of basal cells into differentiated hair follicle cell types. Our data indicate that BCCs likely result from hair follicle progenitor cells that, due to Hh signal activation, cannot progress through normal hair follicle differentiation. These data confirm the role of Ptch1 as a negative regulator of epidermal progenitor turnover and also show for the first time that Ptch1 plays a role in the differentiation of the hair follicle lineage. In addition, we show that insulin-like growth factor binding protein 2 (Igfbp2) is upregulated in both murine and human BCCs and that blocking Igfbp2 activity reduces the Hh-mediated expansion of epidermal progenitor cells. We propose that Igfbp2 mediates epidermal progenitor cell expansion and therefore represents an epidermal progenitor cell–specific target of Hh signaling that promotes BCC development. Cancer Prev Res; 3(10); 1222–34. ©2010 AACR.

Read the Perspective on this article by Peacock and Rudin, p. 1213

Many tumor types have been linked to inappropriate activation of the Hedgehog (Hh) signaling pathway, including brain, prostate, pancreatic, gastric, ovarian, lung, and skin tumors (1–5). It is well accepted that basal cell carcinoma (BCC) skin tumors develop as a result of Hh signal activation, given that the majority of BCCs exhibit mutations in the transmembrane receptor Patched1 (Ptch1), a negative regulator of Hh signaling (6). In the absence of Hh ligand, Ptch1 suppresses the action of a second receptor Smoothened (Smo), thereby maintaining the pathway in a repressed state (7). The binding of Hh ligand to Ptch1 releases Smo and activates downstream signaling, ultimately culminating in the promotion of activator forms of the Gli family of transcription factors, Gli1, Gli2, and Gli3. In the skin, Gli2 functions as the major mediator of Hh signaling (8), supported by the observation that Gli1 activation in the skin primarily functions to increase Gli2 levels (9). There are few conclusive universal targets of Hh signaling, and many targets of the pathway seem to be tissue and context dependent. Cell cycle genes are commonly activated following Hh signal activation (10), and we have previously shown that loss of Ptch1 results in BCC through promoting cell cycle progression (11).

BCCs occur due to the aberrant expansion of epidermal basal cells in the proliferative compartment of the skin. In normal skin, progenitor cells residing in the basal layer of the epidermis proliferate then differentiate upward into the spinous, granular, and cornified layers of the epidermis to create a functional barrier (12–14). Hair follicles begin to arise around 14.5 dpc, with the involution of basal cells into structures referred to as epidermal condensates, which subsequently grow into the underlying dermis, giving rise to the hair follicle outer root sheath (ORS), at the base of which are proliferating epithelial cells known as the hair matrix (15). Signaling between the hair matrix and the adjacent mesenchymal dermal papilla is required to maintain proliferation and induce inward differentiation of cells into inner root sheath (IRS) and hair shaft structures (16). Similar to the interfollicular epidermis (IFE), hair follicles continually regenerate themselves and undergo stages of growth (anagen), regression (catagen), and rest (telogen) in a process termed the hair cycle (17). The dynamics of epidermal cell turnover are critical, and deregulation of these processes result in BCC formation. Pluripotent stem cells are located in the bulge (bulgeSC), a specific region of the ORS located between the permanent and cycling regions of the hair follicle. Bulge cells, which are defined by their expression of keratin (K) 15, are able to regenerate all epidermal lineages (18), although the conditions in which they do so are still being elucidated (19). p63-expressing stem/progenitor cells with regenerative potential are located within the basal layer of the IFE (ifeSC), ORS (bulgeSC), and hair follicle matrix (hfSC; ref. 20). Despite these populations all possessing regenerative and multipotent characteristics, the contribution of these populations to BCC development is unknown.

Various murine models have been used to study the role of Hh signaling in BCC development. In normal skin, Sonic Hh (Shh) ligand production by the dermal papilla signals to the hair matrix to induce the proliferation necessary for hair growth (21–23) and Hh signaling also regulates cell turnover in the IFE (24, 25). Many models of Hh signal activation develop basal cell tumors, including overexpression of Shh, Dhh, Gli1, and Gli2; however, all exhibit differences in tumor latency and phenotypic appearance (24, 26–28). Although the phenotypic variation can be attributed to differences in the timing or level of pathway activation, ultimately these models may not necessarily reflect physiologically relevant levels of signaling. We have previously shown that loss of Ptch1 function in skin results in loss of cell cycle control and the development of BCC, although it remains to be determined whether this is due to direct control of cell cycle genes through Gli or if other intermediate mediators are responsible (11).

Regulation of Hh pathway activity by extrinsic signals and synergistic activity with other signaling networks has been implicated in a number of carcinogenic states; for example, synergism between insulin-like growth factor (Igf) and Hh signaling has been implicated in medulloblastoma (29). However, very few interacting signaling networks have been identified in the initiation and formation of BCC and the Igf family, including the Igf binding proteins (Igfbp), have been implicated in a broad range of cancer types, making the Igf family an interesting target for study in Hh-induced tumorigenesis.

Using K14+ cell–specific deletion of Ptc1 in vivo as the basis of our studies, here we show that loss of Ptc1 activates the proliferation of both IFE and hair follicle progenitor cells followed by the development of BCC-like lesions. We show that Igfbp2 expression is promoted in the absence of Ptch1 and seems to mediate the expansion of the progenitor compartment. We propose that Igfbp2 is a mediator of the effects of Hh signaling on epidermal progenitors and is required to promote BCC development.

The deletion of Ptch1 in the epidermis (K14-Cre:Ptch1^lox/lox) was achieved by crossing female homozygous Ptch1 conditional mice (30) with male mice heterozygous for Ptch1 conditional allele and K14-Cre recombinase (31). The reversed mating (with female K14-Cre carriers) also produced litters with normal Mendelian ratios of offspring. All mice were PCR genotyped and housed in a light-controlled facility, with all work done under the approval of institutional ethics requirements.

RNA was prepared using a QIAgen RNeasy fibrotic tissue column kit (QIAgen) with DNase digestion. cDNA was prepared from 500 ng of RNA, and reverse transcription (RT) was done using M-MLV reverse transcriptase (Invitrogen). Quantitative RT-PCR was done using Applied Biosystems inventoried assays [glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mouse 4352339E, GAPDH rodent 4308313 (discontinued), Igf1 Mm00439561, Igfbp2 Mm00492632m1], and primer and probe sets were designed in house by Karen McCue [Gli1 forward 5′-GCATGGGAACAGAAGGACTTTC-3′, reverse 5′-CCTGGGACCCTGACATAAAGTT-3′, probe 5′-6FAMTCTGCCCTTTTGCCAAGCC-3′ (6FAM); Gli2 forward 5′-CACCTGCATGCTAGAGGCAAA-3′, reverse 5′-AGAAGTCTCCATCTCAGAGGCTCA-3′, probe 5′-TTTTGTCTCGGGTCCGCCA-3′ (6FAM)]. Quantitative RT-PCR was done on ABI 7000 and 7500 RealTime machines (Applied Biosystems), and analysis was done using ABI Prism 7000 SDS software.

Exon 3 deletion of Ptch1 was validated by RT-PCR of exon 2 to exon 6 from cDNA. Amplification of Ptch1 was done using forward (5′-CACCGTAAAGGAGCGTTACCTA-3′) and reverse (5′-TGGTTGTGGGTCTCCTCATATT-3′) primers to amplify fragments of ∼450 bp (wild type) and 250 bp (exon 3 deleted). Whole skin RNA and cDNA were made as above, and PCR amplification was done using a touchdown PCR, with a final annealing temperature of 55°C. Products were separated on 1% agarose gel with Invitrogen 1 kb plus ladder.

Digoxigenin-labeled in situ probes were produced as previously described (32). Fourteen-μm RF paraffin sections were prepared, and in situ labeling was done as previously described (24). Hybridization was done at 64°C with probes to Ptch1, Gli1, and Gli2 (23). Slides were dehydrated and mounted in Entellan mounting medium (Merck).

Staining was done on 4 to 8 μm paraffin sections. Hematoxylin and eosin staining was done as follows: after dehydration, sections were stained for 3.5 minutes with hematoxylin (Vector Labs), washed two times with H₂O, placed in 0.2% (v/v) HCl for 30 seconds, washed again, followed by 10 seconds in LiCl solution. Sections were then taken through ethanol series to 80% and stained with Eosin Y (Sigma) for 30 seconds. Finally, slides were dehydrated in xylene and mounted with Entellan.

Sections were prepared as per histologic stains with citrate buffer unmasking solution as required (Vector Labs), and PBS or TBS buffer was used throughout. Blocking with 1% bovine serum albumin and 4% horse serum in PBS/TBS for 1 hour was followed by incubation with primary antibody overnight at 4°C. Primary antibodies used were as follows. No citrate: AE15 Developmental Studies Hybridoma Bank (DSHB), keratin 6 (Babco), K14 (Covance), K10 (Covance), p63 DSHB. With citrate unmasking: fillaggrin (Babco/Covance), K15 LHK15 (Abcam), Igfbp2 (Santa Cruz), phosphorylated pAkt (CST), phosphorylated p42/44 mitogen-activated protein kinase (MAPK; CST). Secondary antibodies used were as follows: goat α-rabbit Cy3 (Amersham Jackson), sheep α-mouse Cy3 (Amersham Jackson), goat α-mouse Alexa 488 (Invitrogen). Secondary antibodies were used at the manufacturer's recommendation followed by 4′,6-diamidino-2-phenylindole (DAPI) counterstain (Sigma) 1/10,000, and slides were mounted using 1% N-propyl galate in 50% glycerol or fluorescence mounting medium (Dako).

BCC classification was done as part of routine histologic diagnosis after collection. Igfbp2 analysis was done by a single clinical researcher who scored BCC epithelia and stroma as positive or negative for Igfbp2 compared with secondary antibody–alone control. A positive score was classified as the presence of stained cells, regardless of staining pattern, validated on at least duplicate sections. The staining pattern was classified into nuclear or extracellular matrix (ECM)/cytoplasmic staining based on colocalization with DAPI, recorded, and compared with previously classified histologic subtype.

Skin segments (0.5 cm²) were rotor homogenized in 0.5 mL radioimmunoprecipitation assay buffer (50 mmol/L Tris, 150 mmol/L NaCl, 0.5% Na-deoxycholate, 1% NP40, 0.1% SDS, 1× protease inhibitor; Roche) and 1× PhosphoSTOP phosphatase inhibitor (Roche), sonicated, and centrifuged for 30 minutes, 13K, at 4°C, and the supernatant was collected. Quantification was done using the BCA assay (Pierce). All Western blot experiments were done with PageRuler prestained protein ladder (Fermentas).

Twenty-five micrograms of p10 skin protein or 70 μg of adult/p28 skin protein were denatured at 100°C for 5 minutes or 70°C for 10 minutes with 0.1 mol/L DTT then separated on polyacrylamide-SDS gel. All samples were run alongside PageRuler Prestained protein ladder (Fermentas). Protein was transferred onto 0.2-μm nitrocellulose membrane (Whatman, Schleicher and Schuell) or 0.45-μm Immobilon-P polyvinylidene difluoride membrane (Millipore) using 20% methanol, Tris-glycine buffer. Blots were blocked in 5% skim milk in PBS or, for phosphorylated antibodies, 1% bovine serum albumin (Roche) in TBS. All washes were done in PBS/TBS with 0.1% Tween 20 (Sigma). Primary antibody incubations were done overnight at 4°C: p110 phosphoinositide 3-kinase H-239 (Santa Cruz), phosphorylated pAkt (CST), p42/44 MAPK (CST), phosphorylated p42/44 MAPK (CST), total Akt (CST), GAPDH (R&D), and Lef1 (CST). Secondary antibodies used were α-rabbit horseradish peroxidase (Promega) and α-rabbit HRP (CST). Secondary antibodies were used as recommended by the manufacturer for 1 hour at room temperature. Protein was detected through chemiluminescence (Pierce or Amersham) as per the manufacturer's instructions. Densitometry was done at 300 dpi on Quantity One software, and statistical analyses were done on values normalized to control species.

Whole skin from p22 mice was collected and washed through 70% ethanol (two times) and H₂O. Then, 50-mm square biopsies were cut from nonpigmented skin (catagen/telogen hair follicles) and washed two times in penicillin/streptomycin/fungisome (Gibco) containing Ca²⁺/Mg²⁺ free PBS (Gibco), followed by rinsing in Ca²⁺/Mg²⁺–free PBS. Biopsies were then floated on OC medium [DMEM/F12 (1:1), 1× penicillin/streptomycin, 1% l-glutamine, and 1% FCS] for 24 hours before addition of exogenous modulators, including a combination of goat α-mouse Igfbp2 (BD Biosciences) and rat α-mouse Igfbp2-neutralizing antibodies (MAB797), antibody isotype controls (AB108, MAB006), recombinant mouse Igfbp2 protein (797-B2, R&D Systems), or Igf1R inhibitor BMS-536295 (Selleck/Life Research). Explant cultures were collected 2 to 4 days posttreatment, and histology was analyzed. The proportion of pHH3 cells were counted; total DAPI nuclei was calculated from raw count data; and statistical analysis was done as below.

Statistical analyses were done using Word Excel and ABIprism statistical software. Data analysis was as indicated in the figure legend. If not specified, statistical significance was determined using Students t test. In general, the sample replicate average was subtracted from reference replicate average, and the combined SD or SEM of normalized values was calculated.

Basal cell–specific deletion of Ptch1 in the epidermis was achieved by breeding K14-Cre Recombinase mice (31) with a Ptch1 conditional line (30). Ptch1 deletion was validated in the skin by RT-PCR of the recombined region of Ptch1 (Supplementary Fig. S1). By crossing to Z/AP reporter mice (33), we observed that the K14-Cre transgenic line deletes at high levels in the epidermis after birth and only minimally before birth, indicating that this transgene promoter does not completely recapitulate the normal K14 expression pattern observed during development (data not shown). At birth, K14-Cre:Ptch1^lox/lox mice were indistinguishable from control littermates, but rapidly displayed substantial skin defects and severe postnatal growth retardation.¹

They exhibited hair loss and hyperkeratosis, which correlated histologically with a loss of hair follicle structure after the first hair cycle and BCC development, respectively (Fig. 1A). The first discernable histologic defect of K14-Cre:Ptch1^lox/lox skin was expansion of the permanent region of the hair follicle ORS cells and invaginations of the IFE basal cells at around postnatal day 10 (p10; Fig. 1A). This was followed by expansion of the IFE basal cell layer, loss of hair follicle growth associated with the first anagen, and culminated in BCC development by p24-28 (Fig. 1A). Gli1 mRNA upregulation (Fig. 1B) was detected after birth throughout the basal and suprabasal cells of the hair follicle and IFE in K14-Cre:Ptch1^lox/lox epidermis, confirming loss of Ptch1 repression and Hh pathway activation. In support of previous studies indicating that Gli1 likely mediates the Hh signal through Gli2, we also observed Gli2 mRNA upregulation (Fig. 1C) in K14-Cre:Ptch1^lox/lox skin. Increased bromodeoxyuridine labeling of K14-Cre:Ptch1^lox/lox skin (Fig. 1D, g-j), further supported by increased proliferating cell nuclear antigen (PCNA) and cyclin D2 (Fig. 1D, k-n), confirms an increase in proliferation following loss of Ptch1 activity.

K15 is a marker of the hair follicle bulge and thus serves to identify the number and location of the multipotent follicular stem cell (bulgeSC) compartment (18). K15 staining of K14-Cre:Ptch1^lox/lox mice 10 days postbirth remained restricted to the hair follicle region of the skin (Fig. 2A), but revealed a significant expansion in the size of the hair follicle bulge (Fig. 2A). Expression of p63, a progenitor cell marker, was also significantly expanded throughout the entire IFE and upper hair follicle of K14-Cre:Ptch1^lox/lox mice (Fig. 2B). These results indicate an expansion in both the quiescent follicular stem cell compartment and the rapidly proliferating IFE transit-amplifying cells. Expansion of p63 and K15 indicates that Ptch1 is required to negatively regulate both the ifeSC and bulgeSC compartments. The expansion of the stem/progenitor compartment of K14-Cre:Ptch1^lox/lox skin could be attributed to increased proliferation, decreased apoptosis, inhibition of differentiation, or some combination of these processes. As described above, Ptch1 null basal cells exhibit increased proliferation (Fig. 1D), and terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling staining of Ptch1 null skin revealed minimal changes in the rate of apoptosis (Supplementary Fig. S2). We investigated IFE and hair follicle differentiation as a means to clarify the mechanism of progenitor cell expansion. Hair follicle markers Lef1 (Fig. 2C), AE13, and AE15 (data not shown) were decreased in K14-Cre:Ptch1^lox/lox skin, whereas the IFE differentiation markers K10 and filaggrin were increased (Fig. 2D). This suggests that the presence of Ptch1 is required for normal differentiation of the hair follicle lineage but is not necessary for IFE differentiation. Furthermore, these data support the hypothesis that BCCs develop from increased proliferative activity of progenitor cells that cannot undergo normal hair follicle differentiation. On the basis of these data, we can conclude that BCCs likely result from a combination of cell cycle promotion and loss of differentiation following Ptch1 mutation/inactivation.

Apart from skin abnormalities, K14-Cre:Ptch1^lox/lox mice exhibit a severe growth retardation phenotype that we found to be associated with a deficiency in circulating plasma Igf1 levels.¹ Igfbp3, Igfbp5, and Igfbp6 have previously been identified as tissue-specific Hh targets during early embryogenesis or in cultured cells, including keratinocytes (34–38). Given that Igfbps regulate Igf availability and serum half-life, we investigated skin Igfbp levels to identify possible skin-specific mediators of the observed Igf deficiency. Analysis of transcript levels of Igfbp family members in adult K14-Cre:Ptch1^lox/lox skin identified that Igfbp2 is increased after Ptch1 deletion, and hence a possible Hh target in skin (Fig. 3A). Because little is known regarding the role of Igfbp2 in the skin, we then investigated the temporospatial expression of Igfbp2 in control skin. As indicated in Fig. 3, Igfbp2 is normally expressed in a small number of cells at the approximate border of the permanent and cycling hair follicle at p10 (Fig. 3B) and is undetectable in 6-week-old skin (Fig. 3B). Igfbp2 expression is not observed in the more differentiated inner layers of the hair follicle, given the lack of colocalization with K6, which marks the companion cell layer inside the ORS (Fig. 3G). These data indicate that Igfbp2 expression is limited to undifferentiated basal cells in the hair follicle.

K14-Cre:Ptch1^lox/lox skin exhibited increased numbers of cells with apparent elevated levels of Igfbp2 (Fig. 3). In p10 K14-Cre:Ptch1^lox/lox skin, Igfbp2 expression occurs in the ORS, shown by colocalization with K14 (Fig. 3B and C, e-f). We also observed that Igfbp2 localized exclusively to the expanded region of cells surrounding K15⁺ bulgeSCs, but is not coincident with K15 (Fig. 3C, i-j). Igfbp2 expression is localized to the proliferative portion of the hair follicle permanent region of K14-Cre:Ptch1^lox/lox skin, as indicated by PCNA staining (Fig. 3C, k-l). The association of Igfbp2 to the hair follicle proliferative zone, adjacent to the bulge, suggests that Igfbp2⁺ cells possibly represent early, transit-amplifying cells derived from the bulgeSC compartment.

The restricted expression of Igfbp2 in the hair follicle in control skin and its increased expression following Ptch1 deletion indicated that Igfbp2 might mediate Hh-induced hair follicle progenitor cell proliferation. In adult K14-Cre:Ptch1^lox/lox epidermis, Igfbp2 expression is observed throughout the BCC-like lesions, epithelium, and surrounding mesenchyme (Fig. 3). K6-Cre:Ptch1^lox/lox BCC lesions also expressed Igfbp2, further supporting the notion that Igfbp2 is regulated by Ptch1 (Supplementary Fig. S3; ref. 11). These data suggest an association between Igfbp2 expression and BCC development.

To address whether Igfbp2 plays a potential role in human BCC, we screened a collection of human BCC samples for Igfbp2 expression. Twenty of 23 human BCC samples showed an increase in Igfbp2 expression when compared with normal, unaffected skin from surrounding regions (Supplementary Table S1). BCC tumors exhibited two distinct Igfbp2 staining patterns, (a) a nuclear stain (Fig. 4C and F) or (b) a diffuse, cytoplasmic (or possibly ECM) stain (Fig. 4B and E). The remaining three BCC samples (3 of 23) were negative for Igfbp2 expression. We observed no clear correlation between the Igfbp2 staining pattern and BCC subtype (Supplementary Table S1).

To directly address the role of Igfbp2 in the skin, we established an ex vivo organ culture system. In both control and K14-Cre:Ptch1^lox/lox explant cultures, Igfbp2 inhibition using a neutralizing antibody (39) decreased the number of cells expressing high levels of K15 (Fig. 5A), an effect that was particularly pronounced in K14-Cre:Ptch1^lox/lox explant cultures. These data suggest that Igfbp2 regulates the size of the K15⁺ population. p63 staining showed that Igfbp2 inhibition also decreases the expansion of the progenitor compartment that normally occurs in K14-Cre:Ptch1^lox/lox skin (Fig. 5B). K14-Cre:Ptch1^lox/lox explant cultures showed significant decreases in hair follicle proliferation, reflecting the loss of anagen progression seen in vivo (Fig. 1). Anti-Igfbp2 treatment of K14-Cre:Ptch1^lox/lox explant cultures showed a trend (P = 0.015) toward increasing proliferation in the hair follicle (Fig. 5C). Thus, although Igfbp2 has been shown to both inhibit and promote proliferation (40), our results indicate that Igfbp2 may function to mediate hair follicle progenitor cell progression into anagen.

Igfbp2 has been suggested to both activate and inhibit Igf1R signaling (41, 42). To determine whether Hh regulation of Igfbp2 was influencing Igf signaling in the skin, we investigated Igf1R expression and its downstream signaling components. Although we failed to detect phosphorylated Igf1R in either control or K14-Cre:Ptch1^lox/lox skin (data not shown), we observed an increase of plasma membrane–localized Igf1R in the K14-Cre:Ptch1^lox/lox epidermis, indicating potentially altered Igf1R function (Fig. 6B). We also observed a modest decrease in activation levels of the Igf1R signal mediators p42/44 MAPK (a Ras signal effector) and Akt (a PI3K signal effector) in adult K14-Cre:Ptch1^lox/lox epidermis compared with controls (Fig. 6A). p42/44 MAPK and Akt both localized to the differentiating epidermis (Fig. 6C) and have previously been suggested to regulate epidermal differentiation (43, 44). It is therefore possible that the decrease in p42/44 MAPK and Akt observed in K14-Cre:Ptch1^lox/lox skin may reflect a decrease in the proportion of differentiating epidermis. Although p85 and p110 PI3K also mediate Igf1R signaling, their level and localization were not altered following Ptch1 deletion (Fig. 6A and data not shown). p38 MAPK, a recently identified Igf1R signal mediator, was expressed throughout the K14-Cre:Ptch1^lox/lox skin, but was not altered in an Igfbp2-associated pattern (data not shown) and is therefore is not likely to be affected by Igfbp2 in this system. Finally, in explant culture, IgfR1 inhibition using BMS-536924 did not modulate the number of p63 cell layers or the size of the K15-positive region (Fig. 6D) in either control or K14-Cre:Ptch1^lox/lox explants, further suggesting that Igf1R signaling is not the mechanism through which Igfbp2 regulates progenitor cell expansion in Ptch1 null epidermis. Overall, Hh-induced Igfbp2 expression does not seem to function by modulating Igf1R signaling in K14-Cre:Ptch1^lox/lox skin.

Ablation of Ptch1 activity in the K14-Cre:Ptch1^lox/lox mouse model indicates that Hh signal activation promotes the expansion of progenitor cells, in agreement with our previous data (11) and those of others (25). Specifically, we have shown here that upon loss of Ptch1 activity, the hair follicle stem/progenitor compartment expands as a result of increased proliferation and a decrease in hair follicle lineage commitment/differentiation. The increased levels of p63 and K15 we observed in K14-Cre:Ptch1^lox/lox adult epidermis supports the view that BCC develop due to the maintenance of cells in a transit-amplifying (progenitor), non–hair follicle differentiated state. Hh signal downregulation is required for hair follicle regression, a process most likely regulated by Ptch1 (22). Our data indicate that downregulation of Hh activity by Ptch1 is also required to enable cells to stop proliferating and commit to hair follicle differentiation. K14-Cre:Ptch1^lox/lox Ptch1 null epidermis exhibits decreased levels of differentiated hair follicle markers (Lef1 and AE13) despite increased numbers of progenitor cells. This observation suggests that K14-Cre:Ptch1^lox/lox BCC formation is likely due to cells being maintained in a proliferating, nondifferentiated state.

Contrary to the role of Ptch1 in hair follicle differentiation, Ptch1 does not seem to be required for interfollicular differentiation. The expanded region of interfollicular differentiation present in K14-Cre:Ptch1^lox/lox skin likely reflects the increased number of cells, all of which are progressing into the differentiation program.

Igfbp2 is a novel Hh target, and these data clearly implicate Igfbp2 in BCC development. Igfbps have been identified as Hh targets in a number of other tissues and systems, in a tissue- and context-specific manner (34–38). In K14-Cre:Ptch1^lox/lox skin, Ptch1 deletion results in increased Igfbp2 expression specifically in the region of the hair follicle surrounding the bulge, and therefore represents a cell type and possibly context-specific epidermal Hh target. Igfbps have distinct, spatially restricted expression patterns, but regulate a wide range of functions. Igfbp2 may regulate synergism between Hh and Igf hormone signaling specifically in the bulge niche or alternatively mediate interactions of Hh responding cells with their niche through integrin signaling (45, 46).

This study suggests that Igfbp2 promotes epidermal progenitor cell turnover leading to Hh-induced BCC. Treatment with anti-Igfbp2 antibody in explant culture decreases K15 and p63 expression, implying that Igfbp2 positively regulates the progenitor compartment. We propose that following the loss of Ptch1, Igfbp2 activity maintains a proportion of cells in the ORS transit-amplifying compartment in an undifferentiated state. Hair follicle proliferation is significantly lower in Ptch1-deleted explants compared with controls, indicating that anagen cannot proceed in Ptch1-deleted explants, reminiscent of our in vivo observations of hair loss in K14-Cre:Ptch1^lox/lox skin (Fig. 1). Anti-Igfbp2 treatment of Ptch1-deleted explants shows an increased trend in hair follicle proliferation, which we hypothesize to be due to Igfbp2 inhibition allowing ORS progenitor cells to progress into anagen. The overall effect of Igfbp2 antibody treatment on proliferation over the period of explant culture is minimal, suggesting that Igfbp2 may be indirectly inhibiting hair follicle proliferation and that the primary role of Igfbp2 is to maintain hair follicle progenitor cells. Previous studies in other systems lend support to this notion. It has been shown that Igfbp2 maintains pluripotency in hematopoietic stem cells (40), and a recent in vitro study suggests that Igfbp2 maintains p63⁺ main human epidermal progenitor cells (47). Igfbp2 is also clearly associated with BCC and therefore may play a role in maintaining BCC cells in an undifferentiated, hair follicle state.

The data presented here implicate Igfbp2, regulated by Hh signaling, in BCC and suggest the possibility of similar interactions in other Hh-driven tumors. Figure 7 combines the data from this study into a general model of the cell type–specific interactions between Hh and Igfbp2 leading to BCC development, in which we propose that Igfbp2 promotes the progenitor cell state. We propose that after Ptch1 mutation, inappropriate creation of a pluripotent niche causes the accumulation of progenitor cells that leads to BCC development. In this study, Igfbp2 is suggested to act to promote the development of BCC, and regulation of Igfbps by Hh in general may prove relevant to many different tumor types. Igfbp2 expression has been identified in glioblastoma, prostate, breast, and hematopoetic tumors, all of which have been linked to Hh signal activation (40, 42, 48). These tumor types in particular would be worth investigating with regard to the role of Hh and Igfbp2. Furthermore, because Igfbps are tissue-specific Hh targets, a possibility exists that other Igfbp family members may be involved in tumors of different tissue types.

Igfbp2 has previously been reported to both positively and negatively regulate Igf1R signaling and to function in both Igf ligand-dependent and -independent mechanisms (39, 41, 42, 45, 49, 50). While it remains possible that Hh signaling may regulate Igfbp2 to alter activation of Igf1R, we examined a number of markers of Igf1R activation and have no direct evidence of Hh promoting Igf/Igf1R/Igfbp2 interactions. Although phosphorylated active Akt and p42/44 MAPK are significantly lower in adult K14-Cre:Ptch1^lox/lox skin, their localization in the differentiated layers of the epidermis suggests that this likely represents the decreased proportion of differentiated epidermis in late-stage tumors rather than a difference in Igf/Igf1R signaling per se. The absence of an effect of Igf1R inhibition on p63 and K15 in Ptch1-deleted explants also supports the contention that Igf1R signaling is not involved in Igfbp2-mediated bulge expansion after Ptch1 deletion. Overall, these data suggest that Igfbp2 may function independently of Igf1R. Igfbp2 has previously been shown to function in an Igf-independent manner, resulting in the regulation of tumor-microenvironment interactions with integrins or nuclear binding of p21^Cip/waf. Here, we propose that Igfbp2 may be functioning in BCC in a similar manner (45, 50).

No potential conflicts of interest were disclosed.

Grant Support: National Health and Medical Research Council of Australia and the Australian Cancer Research Foundation.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.