Basal Cell Carcinomas in Mice Overexpressing Sonic Hedgehog

See allHide authors and affiliations

Science  02 May 1997:
Vol. 276, Issue 5313, pp. 817-821
DOI: 10.1126/science.276.5313.817


Mutations in the tumor suppressor gene PATCHED(PTC) are found in human patients with the basal cell nevus syndrome, a disease causing developmental defects and tumors, including basal cell carcinomas. Gene regulatory relationships defined in the fruit fly Drosophila suggest that overproduction of Sonic hedgehog (SHH), the ligand for PTC, will mimic loss of ptcfunction. It is shown here that transgenic mice overexpressing SHH in the skin develop many features of basal cell nevus syndrome, demonstrating that SHH is sufficient to induce basal cell carcinomas in mice. These data suggest that SHH may have a role in human tumorigenesis.

A large body of evidence supports the idea that multiple genetic events are required to transform normal epithelium into benign growths and then into metastatic tumors (1). Some types of tumors rarely show complete progression: For example, basal cell carcinomas (BCCs) of the skin—the most common tumors in Caucasians, with about 750,000 new cases annually in the United States—are generally only locally invasive (2). The lack of a mouse model of BCCs and the difficulty in culturing human BCCs has slowed progress in understanding the mechanisms underlying BCC biology.

Basal cell nevus syndrome (BCNS) is an autosomal dominant disease characterized by developmental defects and a predisposition to certain tumors (3). The most common morphologic abnormalities are skeletal defects such as polydactyly, jaw and rib defects, and spina bifida; the most common tumors are BCCs, medulloblastomas, and meningiomas. The defective gene is ptc (4), a gene on chromosome 9q, first identified in Drosophila as a regulator of embryonic pattern formation. Numerous sporadic BCCs also have 9q loss and ptc mutations, suggesting that many BCCs unrelated to BCNS arise from somatic damage to both copies ofptc (4, 5).

The ptc gene encodes a transmembrane receptor that represses transcription of genes encoding transforming growth factor–β and Wnt class signaling proteins and ptc itself (6). One vertebrate PTC ligand is the secreted protein SHH, which binds to PTC in cultured cells and frog oocytes. The fly homolog of SHH, Hedgehog (HH), is believed to inactivate PTC function, suggesting that HH proteins induce target gene transcription by inactivating their receptor’s function.

The COOH-terminal part of SHH is an autoprotease and cholesterol transferase that cleaves the SHH precursor into two fragments and adds a cholesterol moiety to the NH2-terminal fragment (6). The latter fragment is sufficient for all known signaling events and contains a zinc hydrolase-like domain that may act as a peptidase (6, 7). Thus, in addition to binding to PTC, SHH may cleave an unknown target molecule, although no catalytic activity has yet been detected. In Drosophila,ptc represses its target genes except where ptcfunction is inactivated by Hh, and this relationship appears to be conserved in vertebrates. Excess Hh function has an effect similar to loss of ptc function (8). This genetic relationship means that overexpression of Shh in mouse skin might mimic the loss of ptc function seen in human BCCs.

In normal mice, Shh and ptc RNA accumulate in follicular but not interfollicular skin. Initial expression ofShh and ptc in skin occurs in Hardy stage 1 hair follicles. In 14.5-day postcoital (dpc) skin, Shh RNA accumulates at regularly spaced intervals in the ectoderm. Each spot ofShh signal overlies the mesenchymal condensation of a presumptive follicle (9) (Fig. 1A). High levels of ptc RNA accumulate in each underlying mesenchymal condensation and at slightly lower levels in theShh-expressing ectodermal cells (Fig. 1B), presumably because of induction of ptc transcription by SHH (10).

Figure 1

Expression and phenotypes of normal and transgenic mouse skin. 35S-labeled (red) antisenseShh and ptc probes were used to detect expression patterns by in situ hybridization (32). (A) Focal ectodermal (e) Shh RNA accumulation (red grains) in regularly spaced arrays in a presumptive follicle of 14.5-dpc trunk skin. (B) Accumulation of ptc in both the ectodermal cells (e) where Shh RNA is expressed and the underlying mesenchymal cells (m). (C andD) Sections of 18.5-dpc transgenic (trans) skin growths hybridized with (C) Shh or (D) ptc probes. The highest levels of Shh are in the transgenic epithelium (e), with ptc RNA accumulation in epithelial and neighboring dermal (d) cells. (E through H) Transgenic animals display marked skeletal and skin phenotypes. (E) Transgenic forepaw stained with alizarin red (calcified bone) and alcian blue (cartilage) provides an example of the polydactyly that is frequently observed. There are ectopic centers of cartilage between digits (e), terminal phalangeal bifurcations (b), and a persistent rim of ossification at the distal rim (o) (p, phalanx). (F) An 18.5-dpc transgenic embryo that failed to form dorsal spinal processes. This defect leads to a lack of closure of the spinal canal (arrows) and spina bifida. (G) Skin from a wild-type 18.5-dpc embryonic mouse showing regularly spaced skin folds. (H) Skin from a transgenic mouse showing translucent plaques and a lack of normal folds. (I) Schematic of the K14-Shhtransgene, including the K14 promoter fragment, the β-globin 5′ intron, and the K14 polyadenylation sites (12). The bar in (A) represents 40 μm, for (A) through (D). Skeletal preparations were prepared as in (33).

To examine in vivo the effect of excess SHH signaling, we generated transgenic mice that overexpress SHH specifically in the skin. We fusedShh to the keratin 14 (K14) promoter (Fig. 1I) (11), which drives expression as early as 9.5 dpc in the ectoderm and at later stages in both the follicular and interfollicular epithelium (12). In total, 26 transgenic mice derived from pronuclear injection were examined as embryos or neonates; lines could not be established because of perinatal lethality. In transgenic embryos, high levels of Shh RNA and ptc RNA accumulated in the basal layer of the epidermis, both in the follicular and the interfollicular epithelium (Fig. 1, C and D). The heightenedptc expression confirmed that functional SHH was present and capable of inducing the ptc target gene in epidermal cells. The ptc transcripts were also present in the mesenchyme underlying the ectoderm of transgenics, presumably resulting from movement of SHH into these cells.

The K14-Shh transgenic mice exhibited skeletal and skin abnormalities reminiscent of those seen in BCNS. The most frequent abnormality was polydactyly of both the fore- and hind limbs, some of which had eight digits (Fig. 1E). Each digit looked similar to the normal central digits, as in the chicktalpid 3 mutant (13). Distal phalanges were often missing, giving rise to a shortened but wider limb. Distal cartilage bifurcations, ectopic sites of cartilage formation between the digits, and a distal rim of persistent ossification were apparent. Spina bifida, a failure to close the neural tube, was also frequently observed in the transgenics. This defect always affected the caudal portion of the spine and, in severe mutants, extended to the thoracic spine. Skeletal preparations revealed that spinal processes that normally enclose the spinal cord failed to form dorsally (Fig. 1F) (14). These effects on skeletal development suggest that SHH penetrates internal tissues, as does the mesenchymal ectopic ptc expression.

The K14-Shh transgenic mice had multiple BCC-like epidermal proliferations throughout their skin surface after only the first few days of skin development. Dead perinatal embryos invariably had erosions that destroyed much of the skin surface. The skin lacked normal folds and was translucent and friable (Fig. 1H). Skin histology revealed massive proliferations of cells associated with primordial invaginating hair follicles, which were hyperchromatic but cytologically normal. At 18.5 dpc, the epidermal proliferations often involved most of the epidermal surface (Fig. 2, B and C). In mildly affected embryos, one or two epidermal growths were interspersed with six to eight follicles that appeared normal (Fig.2D). In human BCCs, epidermal cells proliferate and form peripheral “palisades,” a columnar epithelium resembling the basal keratinocyte layer. BCCs lack cell adhesion molecules that normally attach basal cells to the basement membrane zone, resulting in clefts between the basement membrane and tumor (15). These histological features of human BCCs were also found in the epidermal proliferations of K14-Shh transgenics (Fig. 2E).

Figure 2

(A through E) Skin histology of K14-Shh transgenic skin. Hematoxylin-eosin–stained sections of wild-type (wt) and transgenic (trans) mouse skin reveal a marked proliferation of the follicular epithelium. (A) Wild-type 17.5-dpc embryo showing normal keratinization and hair bud formation. Note spacing of hair follicles (HF) and wild-type basal (B), squamous (S), and horny (H) layers of stratifying epithelium. (B) Sagittal sections of transgenic 17.5-dpc trunk skin showing a massive downward growth (Gr) of an apparent hair follicle. Note the large size of the invagination compared to adjacent, apparently wild-type hair follicles (HF). (C) Marked proliferation of cells in a 17.5-day, strongly affected mutant embryo. Note the extensive epidermal growth pattern, which encompasses most of the basal cells (outlined by white arrows). (D) Lower power view of sections from mildly affected skin showing the regular number and spacing of many primary hair follicles (HF) with intermittent follicles forming downward growths (Gr). (E) High-power view of cells showing the artifactual clefting around proliferating cells (C), and peripheral palisading of columnar cells (P), characteristic of BCCs. (F) Paraffin-embedded section of human nodular BCC showing the above characteristics. T, tumor tissue; D, dermis. (Gand H) Photographs of 5-week post-graft skin from B6CBF2 wild-type (G) and transgenic (H) skin grafted onto CB.17scid/scid recipients (20). Note the dramatic reduction in pigmented hair in the graft. (I) Low- and (J) high-power view of haemotoxylin-eosin stain of paraffin-embedded skin showing the ongoing differentiation of BCCs into mature hair follicle epithelium. Note the lack of clefting and the presence of multiple mature hair shafts (arrowheads, HS) and sebaceous glands (S) in each tumor bud. Skin grafts performed as in (27). Bar in (J) represents 32 μm for (E) and (J), 80 μm for (A) through (C), (F), and (I), and 200 μm for (D).

The profile of marker proteins in the murine growths paralleled that of human BCCs. Human BCCs express basal keratins such as K14, do not express markers of differentiating stratified epithelium such as loricrin (16), and produce keratins associated with hyperproliferation, such as keratin 6, in the overlying epidermis (16, 17). Basement membrane proteins such as laminin 5 (Lam5) and bullous pemphigoid antigen 2 (BPAG2) are expressed at reduced levels in BCCs (18). In each of these respects, the K14-Shh–induced skin growths resembled BCCs. At 18.5 dpc, the growths were K14-positive (Fig. 3B) and did not express suprabasal differentiation markers (Fig. 3D). The interfollicular stratifying epithelium appeared to differentiate normally (Fig. 3, C and D), with hyperproliferation in overlying epidermis as revealed by keratin 6 expression (Fig. 3F). Both Lam5 (Fig. 3H) and BPAG2 (Fig. 3J) production was lower in K14-Shh proliferations than in control epidermis, as in BCCs (18). Excess SHH therefore has little effect on stratifying epidermis but causes growth of invaginating hair follicles into BCC-like tumors.

Figure 3

Expression of BCC marker proteins in mouse tumors. Immunologic characterization of 18.5-dpc wild-type (wt) and transgenic (trans) skin. For a normal bright-field image, see Fig. 2A. (A and B) Staining of antibodies to K14 reveals that the epidermal proliferations express basal keratins. (C and D) Loricrin (Lor) shows normal granular staining (G), which is consistent with the normal differentiation pattern of stratified epithelium. (E and F) Expression of keratin 6 (K6) increases in transgenic skin, consistent with hyperproliferation of the epithelium. Note the lack of induced staining in the epidermal growths, consistent with previous studies on human BCCs. (G and H) Antibodies to LAM5 reveal decreased expression in transgenic epithelium. Note the reduction in expression throughout the epithelium (arrows). (I and J) Antibodies to BPAG2 reveal a marked decrease of BPAG2 expression in invaginating epithelium (arrows) compared to basal epithelium (arrowheads). Immunohistochemistry of skin was performed as in (34). Bar in (A) represents 62 μm for (A) through (H) and 25 μm for (I) and (J). Fixed sections were treated with rabbit primary antibodies, followed by anti-rabbit rhodamine-conjugated secondary antibodies, and were visualized with the use of a Bio-Rad confocal microscope. B, basal layer; G, granular layer.

BCCs depend on the surrounding stroma for continued growth. Malignant tumors metastasize upon transplantation into mice, but transplanted BCCs undergo growth arrest or differentiation (19). Donor skin from 18.5-dpc K14-Shh transgenic embryos, marked with black or agouti hair, was transplanted to the dorsum of scidmice with white hair and examined 3, 5, and 10 weeks later (20) (Fig. 2, G to J). In each of 16 control grafts, the epithelium matured normally into wild-type hair follicles (Fig. 2G). In each of eight transgenic grafts, markedly reduced numbers of pigmented hairs were observed (Fig. 2H), showing that Shh can interfere with hair development. Transplanted transgenic skin growths did not enlarge and were partially differentiated. Epithelial growths in five grafts showed signs of mature hair follicle differentiation, including multiple mature hair shafts, sebaceous glands, and cysts in the invagination (Fig. 3, I and J). There was no evidence of metastasis. Thus, as with human BCCs, continued growth of tumors in the transgenics requires the proper tissue context.

The sufficiency of SHH for inducing tumors led us to search for humanShh mutations. A preliminary screen of human tumors revealed a Shh mutation in 1 of 43 BCCs, 1 of 14 medulloblastomas, and 1 of 6 breast carcinomas. These mutations were not detected in blood samples from the same patients (21) nor in blood DNA from 100 normal individuals (Fig. 4, A and B) (22). Remarkably, the three tumors contained the same mutation, a change of His133 to Tyr. Human His133, equivalent to His134 in the mouse sequence, is located on the surface of SHH that may interact with a substrate for the hypothetical peptidase reaction (Fig. 4C) (7). The independent occurrence of the same mutation three times in tumors raises the possibility that human His133 to Tyr is a gain-of-function mutation of Shh. No biochemical assay is yet available for the putative zinc hydrolase activity, but in vivo tests may reveal altered activities of the changed signaling molecule.

Figure 4

Mutations of Shh in human tumor tissues. (A) Direct sequencing from PCR products reveals a mutation in the human Shh sequence at His133 (mouse His134) in BCC tumor tissue but not in the blood of the same individual. The particular nucleotide changed was the same for each tumor type. (B) The mutation was confirmed by restriction fragment length polymorphism. The mutation in tumor tissue (T) eliminates an MscI site in Shh (arrow) present in blood (B) or normal (N) controls. (C) Molecular model of murine SHH (7) reveals the position of mouse His134 (white atoms; equivalent to human His133). Putative catalytic (green) and zinc coordinating (red) residues are located near the changed residue. Molecular modeling was performed with RasMac. M, marker; MED, medulloblastoma; Breast, breast carcinoma.

Normally Shh is expressed in localized areas to control pattern formation. Ubiquitous epidermal expression would reduce or erase the asymmetry, preventing normal digit fates and interfering with formation of dorsal neural tube–derived structures (6,23). In BCNS patients, the haploinsufficiency of ptcmay sensitize anterior limb or dorsal neural tube tissues to normal levels of Shh, resulting in polydactyly or spina bifida. That a twofold change in PTC dose may cause such changes in humans suggests that a precise balance between SHH and PTC is required.

The expression of Shh in basal keratinocytes is sufficient to induce mouse skin tumors that are indistinguishable from human BCCs. This effect of Shh may be completely unrelated to its normal function in skin development; Shh targets induced at high levels could create novel cell types and growth properties. However, the patterns of Shh and ptctranscription in the developing follicle are consistent with a normal role in controlling the proliferation of basal cells in the follicular epithelium. The decrease in mature hairs in the grafts along with the presence of abortive hair follicles in differentiating tumor buds suggests that SHH stimulates the growth of pluripotent follicular epithelium at the expense of differentiation. Such a role is supported by recent studies in chick feather bud development, whereShh induces feather bud outgrowth (24). Juxtaposition to a novel tissue environment appears to inhibitShh activity, resulting in follicular differentiation.

The rapid and frequent appearance of Shh-induced tumors suggests that disruption of the SHH-PTC pathway is sufficient to create BCCs. The mouse BCCs appear within the first 4 days of skin development, unlike mouse squamous neoplasia, where tumors arise 1 to 12 months after oncogene expression (25). The K14-Shh tumor kinetics are consistent with previous clinical and epidemiologic data, which suggest that BCCs, in contrast to melanomas and squamous cell carcinomas, lack precursor or intermediate cellular phenotypes (2).

The gene ptc joins APC in a class of genes instrumental for controlling early epithelial proliferation. Mutations in APC cause familial adenomatous polyposis, a condition that predisposes individuals to many benign polyps, akin to the hundreds of nodular BCCs that can occur in BCNS patients (1). Nodular BCCs are reminiscent of polyps in colonic epithelium, as both lack aneuploidy and are locally invasive (1,2). However, only the initial stages of skin and colon epithelial growth are similar, as colonic polyps progress to metastatic lesions at a substantially greater rate. This could be due in part to the different sets of genes regulated by ptc and APC.

Activating mutations of Shh (or another Hh) may be an alternative pathway for BCC formation in humans. The mutation of human His133 (mouse His134) to Tyr is a candidate. It is distinct from loss-of-function mutations reported for individuals with holoprosencephaly (26). It lies adjacent in the catalytic site to His134 (mouse His135), one of the conserved residues thought to be necessary for catalysis. Our data provide evidence that Shh may be a dominant oncogene in multiple human tumors, a mirror to the tumor suppresser activity of the opposing ptc gene. The mouse model of the disease may be useful in devising strategies for treating the human disease.

  • * To whom correspondence should be addressed. E-mail addresses: scott{at}, ehepstein{at}


Stay Connected to Science

Navigate This Article