Indian hedgehog and β-catenin signaling: Role in the sebaceous lineage of normal and neoplastic mammalian epidermis

Abstract

In mammalian epidermis, the level of β-catenin signaling regulates lineage selection by stem cell progeny. High levels of β-catenin stimulate formation of hair follicles, whereas low levels favor differentiation into interfollicular epidermis and sebocytes. In transgenic mouse epidermis, overexpression of β-catenin leads to formation of hair follicle tumors, whereas overexpression of N-terminally truncated Lef1, which blocks β-catenin signaling, results in spontaneous sebaceous tumors. Accompanying overexpression of β-catenin is up-regulation of Sonic hedgehog (SHH) and its receptor, Patched (PTCH/Ptch). In ΔNLef1 tumors Ptch mRNA is up-regulated in the absence of SHH. We now show that PTCH is up-regulated in both human and mouse sebaceous tumors and is accompanied by overexpression of Indian hedgehog (IHH). In normal sebaceous glands IHH is expressed in differentiated sebocytes and the transcription factor GLI1 is activated in sebocyte progenitors, suggesting a paracrine signaling mechanism. PTCH1 and IHH are up-regulated during human sebocyte differentiation in vitro and inhibition of hedgehog signaling inhibits growth and stimulates differentiation. Overexpression of ΔNLef1 up-regulates IHH and stimulates proliferation of undifferentiated sebocytes. We present a model of the interactions between β-catenin and hedgehog signaling in the epidermis in which SHH promotes proliferation of progenitors of the hair lineages whereas IHH stimulates proliferation of sebocyte precursors.

Mammalian skin is maintained by stem cells whose daughters differentiate along the lineages of the hair follicle, interfollicular epidermis (IFE), and sebaceous gland (1-3). The stem cells are common targets for neoplastic conversion, and the range of different types of epithelial tumor reflects aberrant differentiation along the different epidermal lineages. Thus, whereas basal cell carcinomas (BCCs) may reflect the relatively undifferentiated phenotype of the hair follicle outer root sheath, squamous cell carcinomas exhibit elements of IFE differentiation, sebaceous tumors contain terminally differentiated sebocytes, and pilomatricomas and trichofolliculomas contain cells undergoing differentiation along hair lineages (4).

Many of the molecules that regulate epidermal self-renewal and differentiation have now been identified (1, 2). A key role has emerged for β-catenin signaling in response to the diverse repertoire of Wnts expressed in different regions of the epidermis (5, 6). When β-catenin levels are elevated by expressing a stabilized, N-terminally truncated form of the protein, there is de novo formation of hair follicles in postnatal IFE (7). Conversely, when β-catenin is absent, or its activity blocked with dominant negative forms of the downstream transcription factor Lef1, hair follicles are converted into cysts of IFE with associated sebocytes (8-10). It thus appears that the level of β-catenin controls lineage selection in the skin, with high levels promoting hair follicle formation and low levels stimulating the differentiation of IFE and sebocytes (10, 11).

A second important signaling pathway in the cutaneous epithelium involves the secreted protein Sonic hedgehog (SHH), its receptors Patched (PTCH/Ptch) and Smoothened, and downstream transcription factors of the GLI family (12). Hair follicle development in SHH null mutant mice arrests after the initial epidermal-dermal interactions, indicating that SHH signaling is required for normal advancement beyond the hair germ stage of development (13, 14). Mice that are homozygous null for Indian hedgehog (IHH) have major skeletal defects and die at birth, but no analysis of their skin has been described (15). GLI2 is the key mediator of SHH responses in skin and positively regulates GLI1; GLI2 knockout mice show a similar arrest in hair follicle development to SHH null animals (16). In adult mouse skin SHH is required for the normal hair growth cycle (17), and treatment with antibodies to SHH blocks the growth of anagen hair follicles (18). In both the knockouts and the antibody-treated epidermis some markers of follicular differentiation are expressed, and it seems that SHH is primarily required for growth rather than differentiation (16, 19, 20).

Both the β-catenin and hedgehog (Hh) pathways play a role in epidermal carcinogenesis. PTCH is mutated in human nevoid BCC syndrome (Gorlin-Goltz's syndrome), a hereditary predisposition to BCCs, medulloblastomas, and rhabdomyosarcomas (21-24). PTCH/Ptch mutation results in overexpression and activation of GLI1, and there is evidence that activation of GLI1 and GLI2 leads to the development of BCCs (25, 26). Constitutive activation of SHH signaling in transgenic mouse and human skin also leads to the formation of BCCs (27-29). Smoothened is required for activating transcription of Hh target genes, and overexpression of constitutively active Smoothened results in tumors similar to those caused by SHH overexpression (30).

Transgenic mice that overexpress N-terminally truncated β-catenin in the basal layer of the IFE and hair follicle develop hair tumors (trichofolliculomas and pilomatricomas), and activating mutations in β-catenin have been found at high frequency in human pilomatricomas (7, 31). Activation of β-catenin signaling is associated with cancer in many different tissues, whereas inhibition is not thought to play any role in the disease (32). It is thus surprising that mice expressing N-terminally truncated Lef1 (ΔNLef1), which blocks β-catenin signaling in the basal layer of the IFE and hair follicle (K14ΔNLef1 transgenics), develop spontaneous tumors (10). Consistent with the role of β-catenin levels in controlling lineage selection, the tumors in K14ΔNLef1 transgenics show sebaceous and squamous differentiation rather than hair follicle differentiation (10).

There is good evidence for crosstalk and conservation in Hh and Wnt signaling. Both pathways share components such as GSK, CK1, and the F box protein Slimb (33, 34) and involve G protein-coupled receptors (Smoothened in the Hh pathway and Frizzled receptors for Wnts). When β-catenin is overexpressed in the epidermis, SHH transcription is elevated (7). Conversely, SHH expression is inhibited in β-catenin null epidermis (8), and overexpression of the Wnt antagonist Dickkopf1 blocks SHH expression (35). However, SHH is not simply downstream of Wnt signaling because Wnt5a is a target of SHH in hair follicle morphogenesis (6) and Ptch mRNA is up-regulated in the spontaneous tumors of K14ΔNLef1 transgenic mice (10). In addition, Wnt expression is altered by activation of SHH in human BCCs (36).

To investigate the mechanism leading to development of spontaneous tumors in K14ΔNLef1 transgenic animals we analyzed components of the Hh signaling pathway in more detail. Whereas SHH has been shown to be important for hair follicle development and BCC formation, our results indicate that IHH is involved in growth and differentiation of sebocytes in normal skin and in the formation of sebaceous tumors of human and mice. We propose that ΔNLef1 and IHH cooperate to control proliferation and differentiation of sebocyte progenitors.

Materials and Methods

Source of Tissues. Spontaneous skin tumors from K14ΔNLef1 transgenic mice (10) were formalin-fixed and paraffin-embedded. Formalin-fixed and paraffin-embedded human skin tumors (nine sebaceous adenomas and one pilomatricoma) were obtained from Beth Israel Deaconess Medical Center. Sections from normal mouse and human skin were obtained from the Center for Nutrition and Toxicology, Karolinska Institute, Stockholm and the Department of Dermatology, Karolinska Hospital, Stockholm, respectively.

Immunohistochemical Staining of Tumors and Normal Skin. Paraffin sections were dewaxed, rehydrated, and incubated in 1% H₂O₂ in methanol for 30 min. The sections were then treated with 10 mM citrate buffer at 97°C for 30 min. Sections were incubated overnight at 4°C with diluted rabbit polyclonal antibodies directed against human PTCH1 (Research Genetics, Huntsville, AL, and Santa Cruz Biotechnology, sc-6149, both 1:200), GLI1 (AbCam, Cambridge, U.K., 1:300), GLI2 (Research Genetics, 1:50), and pan-HH (Santa Cruz Biotechnology, sc-9024,1:100) or goat polyclonal antibodies directed against human SHH (Santa Cruz Biotechnology, sc-1194, 1:100), IHH (Santa Cruz Biotechnology, sc-1196, 1:30), and Desert Hh (DHH) (Santa Cruz Biotechnology, sc-1193, 1:100). All antibodies were diluted in PBS containing 0.1% BSA. Detection was carried out with the Vectastain Elite Kit (Vector Laboratories) by using rabbit or goat IgG. After washing the sections with PBS, biotinylated secondary antibodies were added for 30 min at room temperature. After extensive rinsing and incubation with avidin-biotin, immunoperoxidase antibody staining was visualized with 3-amino-9-ethylcarbazole (Vector Laboratories), and sections were counterstained with Mayer's hematoxylin. As controls primary antibodies were omitted (in the case of PTCH1 N terminus, IHH, DHH, and pan-Hh antibodies) or incubated with the corresponding peptide immunogen (PTCH C terminus, SHH, GLI1, and GLI2 antibodies) during the staining procedure.

Sebocyte Culture and Retroviral Infection. SZ95 are a line of human facial skin sebocytes that have been immortalized by transfection of simian virus 40 large T antigen (37) and were cultured in Sebomed medium (Biochrom, Berlin) containing 10% FCS (Sera-Lab, Crawley Down, Sussex, U.K.), 3 ng/ml keratinocyte growth factor (PeproTech, Rocky Hill, NJ), and 20 ng/ml epidermal growth factor (PeproTech). SZ95 cells were retrovirally infected by using supernatant of AM12 amphotropic producer cells as described (38) and selected in 1.5 μg/ml puromycin. To inhibit Hh signaling, subconfluent and undifferentiated SZ95 cultures were treated with 5 μM cyclopamine in methanol; control cells were treated with methanol alone.

To generate growth curves, the Cyto Tox assay kit (Promega) was used to determine total cell number. A total of 2 × 10³ SZ95 sebocytes were seeded per well into 96-well plates, and the number of cells per well was determined in triplicate at different time points for up to 9 days of cell culture.

Staining of SZ95 Cells and Epidermal Whole Mounts. SZ95 sebocytes grown on coverslips were fixed in 4% paraformaldehyde and subjected to indirect immunofluorescence staining as described (38). Primary antibodies used were anti-PTCH antibody (C terminus, Research Genetics, 1:200), anti-IHH antibody (Santa Cruz Biotechnology, 1:200), anti-pan-Hh antibody (Santa Cruz Biotechnology, sc-9024, 1:200), anti-GLI1 antibody (AbCam, 1:300), anti-GLI2 antibody (Research Genetics, 1:50), FWCAD (rabbit anti-E-cadherin antibody; ref. 39), 9E10 (anti-Myc-tag antibody, 1:400), and 12CA5 (antihemagglutinin tag; ref. 40). Secondary antibodies were conjugated with AlexaFluor 488 or AlexaFluor 594 (Molecular Probes). In some experiments cells were counterstained with 4′,6-diamidino-2-phenylindole (Molecular Probes) to identify nuclei. To stain lipids, SZ95 sebocytes were incubated with Nile red (0.1 μg/ml in PBS, Sigma) for 30 min at room temperature, then washed several times with PBS and distilled water.

Whole mounts of tail epidermis were prepared from WT and K14ΔNLef1 transgenic mice bred in a CBAxC57Bl6 F₁ mixed background (10) by using a recently developed method (70). The tail was slit length-wise, and the skin was peeled from the tail. The skin was then incubated in EDTA until the epidermis could be peeled as an intact sheet from the dermis. The epidermis was fixed in 4% formal saline (Sigma) for 2 h at room temperature, then blocked and permeabilized in phosphate buffer consisting of 0.5% skim milk powder, 0.25% fish skin gelatin (Sigma), and 0.5% Triton X-100 in TBS (0.9% NaCl/20 mM Hepes, pH 7.2). Epidermal sheets were incubated with primary antibodies diluted in phosphate buffer overnight at room temperature with gentle agitation. Primary antibodies used were anti-GLI1 antibody (AbCam, 1:300), anti-GLI2 antibody (Research Genetics, 1:50), and anti-β₁ integrin antibody (MB 1.2, kindly provided by B. Chan, University of Western Ontario, London, Canada; ref. 41). The epidermis was then washed for >4 h with PBS containing 0.2% Tween 20, changing the buffer several times. Incubation steps with secondary antibodies were performed in the same way, and epidermal sheets were rinsed in distilled water before mounting in Gelvatol (Monsanto, St. Louis) containing 2.5% 1,4-diazabicyclo[2.2.2.]octane (Sigma).

Images of SZ95 sebocytes and epidermal whole mounts were acquired by using a Zeiss 510 confocal microscope. Scans of whole mounts are presented as z-projections of ≈30 optical sections captured.

Western Blotting. Protein lysates for Western blotting were extracted from undifferentiated, differentiated, and retrovirally infected cultures of SZ95 sebocytes. Cells were washed twice with ice-cold PBS and TNE buffer containing 40 mM Tris·HCl (pH 7.4), 10 mM EDTA, and 150 mM NaCl was added to harvest the cells. SZ95 sebocytes were washed with ice-cold PBS and lysed in 200 μl WCE buffer containing 20 mM Hepes (pH 7.9), 0.42 M NaCl, 0.5% Nonidet P-40, 0.5% deoxycholic acid, 25% glycerol, 0.2 mM EDTA, 1.5 mM MgCl₂, 1 mM DTT, and protease inhibitors. Protein lysates were incubated at 4°C for 30 min with constant agitation and centrifuged at 4°C for 30 min at 14,000 rpm. Proteins in the supernatant were separated on 7.5% (PTCH) and 15% (Hh) SDS/PAGE gels, electroblotted, and probed with antibodies against PTCH (C terminus, 1:200 dilution, Research Genetics), pan-Hh (Santa Cruz Biotechnology, 1:250), Myc-epitope (9E10, Santa Cruz Biotechnology), and hemagglutinin tag (CAS12; ref. 40) as described (38). Immunoreactive proteins were visualized by chemiluminescence (ECL, Amersham Pharmacia). Blots were stripped with glycine buffer and reprobed with an antiactin antibody (AC-40, Sigma, 1;2000) as a control for equal loading.

Results

IHH Is Expressed in Mouse and Human Sebaceous Tumors. Transgenic mice (K14ΔNLef1) expressing a truncated mutant form of the transcription factor Lef1 (lacking the first 32 aa that contain the β-catenin binding domain) under control of the keratin 14 promoter develop spontaneous skin tumors (10). The majority of the tumors are sebaceous adenomas, consisting of differentiated sebocytes surrounded by undifferentiated progenitor cells. The second most common tumor type contains areas of sebaceous differentiation and areas of squamous differentiation or undifferentiated basal keratinocytes (sebeomas; ref. 10). Some of the mice develop squamous papillomas, but even though IFE differentiation predominates in these tumors there are also clusters of sebocytes.

We showed previously by in situ hybridization that ptch RNA was up-regulated in all of the tumors of K14ΔNLef1 mice (10). To investigate the underlying mechanism, we analyzed expression of various components of the Hh signaling cascade in more detail. We performed immunohistochemical staining with antibodies against PTCH, its three known ligands (SHH, DHH, and IHH), and antibodies to downstream transcription factors of the GLI family, GLI1 and GLI2, that are known to be expressed in the epidermis (ref. 16 and Table 1). We examined 12 tumors from K14ΔNLef1 mice, comprising sebaceous adenomas, sebeomas, and squamous papillomas with sebocyte differentiation (Fig. 1 and Table 1).

Table 1. Expression of components in the SHH-Ptch signaling pathway in spontaneous sebaceous skin tumors of K14ΔNLef1 transgenic mice.

		PTCH1
No.	Tumor	C terminal	N terminal	SHH	IHH	DHH	pan-HH	GLI1	GLI2
1	Pap	++	+	+	++	−	+	+	+
2	Seb	+	+	+	++	−	++	+	++
3	SA	+	+	−	++	−	+	+	++
4	Seb	+	+	−	++	−	++	+	+
5	SA	++	+	+	++	−	++	+	+
6	SA	++	+	+	++	−	++	+	++
7	SA	++	+	+	++	−	+	+	++
8	SA	+	+	−	++	−	++	+	++
9	SA	++	+	−	++	−	++	+	++
10	SA	++	+	+	++	−	+	+	++
11	Pap	++	+	+	++	−	++	+	++
12	Seb	+	+	−	++	−	+	+	+

pap, squamous papilloma with sebocyte differentiation; seb, sebeoma; SA, sebaceous adenoma. Expression strength was scored as follows: −, no detectable specific immunoreactivity; +, specific immunoreactivity in a moderate number of cells; ++, strong and specific immunoreactivity in a high number of cells.

Fig. 1.

Expression of proteins in the PTCH-SHH signaling pathway in K14ΔNLef1 transgenic mouse sebaceous tumors. Immunostaining was performed with antibodies to the proteins indicated. (a) Ptch1 C terminus. (b) Ptch1 N terminus. (c) SHH. (d) IHH. (e) DHH. (f) pan-HH. (g) GLI1 (h) GLI2. (Scale bar: 200 μm.)

Staining with antibodies to the N- and C-terminal regions of Ptch1 established that the protein was up-regulated. The Ptch1 C-terminal antibody showed prominent immunoreactivity in both the differentiated sebocytes and the surrounding basaloid cells, labeling the cytoplasm, nucleus, and cell membrane (Fig. 1a). In contrast, the Ptch1 N-terminal antibody stained the cytoplasm and plasma membrane, but gave no or weak nuclear staining (Fig. 1b). These observations allow us to draw two conclusions: first, that the elevated ptch mRNA in the tumors (10) correlates with elevated PTCH protein; second, that Ptch1 is unlikely to be mutated, because the majority of PTCH mutations in tumors result in premature termination of Ptch translation leading to the absence of the C terminus (42).

The expression of SHH protein was generally weak in spontaneous mouse sebaceous tumors. The strongest immunoreactivity with the SHH antibody was seen in the basaloid progenitor cells with only faint staining in the mature sebocytes (Fig. 1c and Table 1). Immunohistochemical staining with the DHH antibody was negative in both the undifferentiated and differentiated cells of the tumors (Fig. 1e and Table 1). In contrast, IHH protein was strongly expressed in the mature sebocytes of the tumors, whereas the basaloid progenitor cells were completely negative in all cases (Fig. 1d and Table 1). Staining with the pan-Hh antibody was seen both in basaloid and sebaceous gland cells and corresponded to the combination of the staining pattern for SHH and IHH in the skin tumors (Fig. 1f). GLI1 and GLI2 proteins were expressed in the basaloid cells and in the sebocytes of the spontaneous skin tumors of K14ΔNLef1 mice, although they were absent from the most highly differentiated sebocytes in the centers of the tumors (Fig. 1 g and h and Table 1). We conclude that the major Ptch ligand expressed in the sebaceous tumors is IHH and that Hh signaling is active because Ptch1 is expressed. The nuclear localization of GLI1 is a further indication that Hh signaling is activated in the sebaceous tumors.

We next examined a panel of nine human sebaceous adenomas. The immunohistochemical staining of components in the Hh-PTCH1 signaling pathway was similar to that observed in the spontaneous mouse tumors (Table 2 and Fig. 2). All of the human tumors stained positive with antibodies to the N and C termini of PTCH1 (Fig. 2 a and b). Strong staining with antibodies against IHH was detected in the mature sebocytes of the human tumors (Fig. 2d), whereas DHH protein was either not detectable (Fig. 2e) or only weakly positive (Table 2) and SHH staining was weak (Fig. 2c). Staining with the pan-HL antibody corresponded to the combined staining patterns for the individual Hh proteins (Fig. 2 f). GLI1 and GLI2 were located in the nucleus and in the cytoplasm (Fig. 2 g and h), although GLI1 antibodies stained the cytoplasm more strongly than the nucleus (Fig. 2g).

Table 2. Expression of components in the SHH-PTCH signaling pathway in sebaceous human skin tumors.

		PTCH 1
No.	Tumor	C terminal	N terminal	SHH	IHH	DHH	pan-HH	GLI1	GLI2
1	SA	++	+	+	++	−	++	nd	+
2	SA	++	+	−	++	−	++	nd	+
3	SA	++	+	+	++	+	++	nd	−
4	SA	++	+	−	++	−	++	nd	+
5	SA	++	+	+	++	+	++	++	++
6	SA	++	+	+	++	−	++	++	+
7	SA	++	+	+	++	−	++	++	+
8	SA	++	+	+	++	−	++	++	+
9	SA	++	+	+	++	−	++	++	+

SA, sebaceous adenoma; nd, not determined. Expression strength was scored as follows: -, no detectable specific immunoreactivity; +, specific immunoreactivity in a moderate number of cells; ++, strong and specific immunoreactivity in a high number of cells.

Fig. 2.

Expression of proteins in the SHH-PTCH signaling pathway in human sebaceous tumors. Immunostaining was performed with antibodies to the proteins indicated. (a) PTCH1 C terminus. (b) PTCH1 N terminus. (c) SHH. (d) IHH. (e) DHH. (f) pan-HH. (g) GLI1. (h) GLI2. (Scale bar: 200 μm.)

Taken together, our results show that components of the Hh-PTCH1 signaling pathway are up-regulated in sebaceous skin tumors of human and mouse, suggesting that the pathway is active in the tumor cells. IHH is the most prominent ligand of the Hh pathway in sebaceous tumors. In contrast, IHH protein was not detectable in human pilomatricomas, which are hair follicle tumors (Fig. 3b).

Fig. 3.

Expression of IHH, Gli1, and Gli2 proteins in normal and K14ΔNLef1 transgenic mouse skin and in human pilomatricoma. (a) Immunostaining with IHH antibody in normal mouse skin. Note strong reactivity in the sebaceous glands (arrowheads). (b) In human pilomatricoma (a hair follicle-derived tumor) immunostaining for IHH was not detected. (c-f) Immunostaining for GLI1 (green, c and d) and GLI2 (green, e and f) proteins in whole mounts of tail epidermis of WT (c and e) and K14ΔNLef1 transgenic (d and f) mice. Note strong expression of GLI1 in the nuclei of cells in the normal sebaceous glands (arrowheads in c) and newly differentiating sebocytes along the deformed transgenic hair follicles (arrowheads in d). Note strong and predominantly cytoplasmic staining of GLI2 in the permanent portion of the follicle below the sebaceous glands in WT epidermis (e, brackets) and weaker, more uniform staining in transgenic epidermis (f, bracket indicates equivalent region to those shown in e). Note weak nuclear staining for GLI2 in normal sebaceous glands (e, arrowheads). Whole mounts were double-labeled for β1 integrins (red) to reveal the basal IFE, periphery of sebaceous glands and outer root sheath of the hair follicles. [Scale bars: 100 μm (a and b) and 50 μm (c-f).]

IHH and GLI1 Expression in Normal Epidermis. To investigate whether the strong expression of IHH in sebaceous tumors was indicative of normal sebaceous differentiation, we analyzed expression of IHH, GLI1, and GLI2 protein in normal skin (Fig. 3). In both human and mouse skin, IHH protein was strongly expressed in the differentiated sebocytes of sebaceous glands and undetectable in the IFE or the hair follicles (Fig. 3a and data not shown).

To examine the expression of GLI1 and GLI2 we stained whole mounts of mouse tail skin (70). We observed weak cytoplasmic immunostaining with antibodies against GLI1 in the outer root sheath of the hair follicle and in the IFE, consistent with a previous study (43). In contrast, strong staining with the anti-GLI1 antibody was seen in the nuclei of progenitor cells in the sebaceous gland (Fig. 3c, arrowheads). Furthermore, we observed nuclear GLI1 protein in sebocyte progenitors that formed ectopically along the deformed hair follicle structures in K14ΔNLef1 transgenic mice (Fig. 3d, arrowheads, and ref. 10). GLI2 was strongly expressed in the cytoplasm of cells in the permanent portion of the follicle below the sebaceous glands (a region where the follicle stem cells are concentrated; ref. 70) and had a nuclear distribution in some cells at the periphery of the sebaceous glands (Fig. 3e, arrowheads). In K14ΔNLef1 epidermis staining for GLI2 was uniform along the deformed follicles and was predominantly cytoplasmic with some weak nuclear staining (Fig. 3f). Taken together, our results suggest that IHH and GLI1 are expressed in normal sebocytes, the ectopic sebocytes induced by ΔNLef1, and sebaceous skin tumors of mouse and human. The distribution of GLI2 in normal follicles is consistent with its primary function in SHH signaling (16). IHH Regulates Growth of Human Sebocytes in Vitro. To investigate the role of IHH signaling in sebocyte differentiation we used the SZ95 line of human facial skin sebocytes that have been immortalized by transfection of simian virus 40 large T antigen (37). SZ95 sebocytes can be induced to differentiate into mature sebocytes by growing the cells to high density (37). Nile red is a hydrophobic dye that is highly selective for lipids and can therefore be used as a marker for differentiated mature sebocytes (ref. 37 and Fig. 4 a and b).

Fig. 4.

Expression of IHH, PTCH1, GLI1, and GLI2 in undifferentiated and differentiated human SZ95 sebocytes in vitro. (a, c, and e) Undifferentiated cells. (b, d, and f) Postconfluent, differentiated cells. (g-i) Partially differentiated cultures. (a and b) Nile red staining of lipids. (b) Note dramatic increase in the number of lipid droplets in differentiated SZ95 sebocytes (arrows). (c and d) Immunostaining for IHH protein (green) and E-cadherin (to reveal cell-cell borders; red). (d) IHH protein levels are up-regulated in differentiated sebocytes (arrows). (e and f) Immunostaining for PTCH1 protein (green) with 4′,6-diamidino-2-phenylindole nuclear counterstain (blue). PTCH1 staining is up-regulated in the cytoplasm of differentiated sebocytes (arrows in f). Immunostaining for GLI1 (g and h) and GLI2 (i) (green) with 4′,6-diamidino-2-phenylindole nuclear counterstain (blue), showing nuclear and cytoplasmic GLI localization in areas where the cultures are still undifferentiated (g and i) and exclusively cytoplasmic localization in suprabasal, terminally differentiating sebocytes (h and i Inset, arrows). [Scale bar: 20 μm(a-i).] (j) Western blots of protein lysates prepared from undifferentiated and differentiated SZ95 sebocytes (Left and Center) and retrovirally infected differentiated cells (Right) probed with antibodies against PTCH1 C terminus (160 kDa), pan-Hh (processed form of ≈25 kDa) and, as a loading control, actin (42 kDa). ev, empty vector control; ΔNLef1, retroviral vector.

First, we compared expression of IHH and PTCH1 in undifferentiated and differentiated SZ95 sebocytes by immunofluorescense staining. IHH was only weakly expressed in undifferentiated cells (Fig. 4c). In contrast, IHH could be readily detected in the cytoplasm in close approximation to the plasma membrane of differentiated mature sebocytes (Fig. 4d). A similar staining pattern was observed with anti-PTCH1 antibodies: weak expression in undifferentiated cells (Fig. 4e) and up-regulated levels in differentiated sebocytes (Fig. 4f). PTCH1 was detected only in the cytoplasm and not in the nucleus of the sebocytes (Fig. 4f). The immunofluorescence results were confirmed by Western blotting with antibodies to PTCH1 and a pan-Hh antibody: levels of both PTCH1 and Hh were higher in differentiated SZ95 cells than in undifferentiated cells (Fig. 4j Left and Center). These data indicate that IHH and its receptor, PTCH1, are up-regulated during sebocyte differentiation in vitro.

The observation that IHH was expressed in mature sebocytes and that GLI1 was activated in sebocyte progenitors suggested a paracrine effect of IHH. This finding was supported by examining GLI1 and GLI2 expression in SZ95 sebocytes (Fig. 4 g-i). In newly confluent cultures both transcription factors were found in the nucleus and cytoplasm of cells that had not yet accumulated substantial numbers of lipid droplets (Fig. 4 g and i). Suprabasal and differentiated cells had cytoplasmic but not nuclear GLI1 (Fig. 4h, arrows) and GLI2 (Fig. 4i Inset, arrow).

To investigate the role of IHH signaling in growth and differentiation of human sebocytes in vitro, we treated SZ95 sebocytes with cyclopamine, a specific and potent inhibitor of Hh signaling (44-46). Treatment of SZ95 sebocytes with 5 μM cyclopamine decreased proliferation (Fig. 5a) and stimulated differentiation, as evaluated by Nile red staining (Fig. 5 b and c). From this we conclude that IHH promotes the proliferation of undifferentiated sebocytes.

Fig. 5.

Manipulating Hh and β-catenin signaling regulates growth and differentiation of human sebocytes in vitro.(a-c) Inhibition of Hh signaling in human sebocytes decreases proliferation and stimulates differentiation. (a) Growth curve of SZ95 cells treated with Hh inhibitor cyclopamine or solvent methanol. (b) Treatment of undifferentiated SZ95 cells with methanol did not affect differentiation as evaluated by Nile red staining, whereas treatment with cyclopamine (c) resulted in accumulation of Nile red-positive lipid droplets (arrows). (d-j) Retroviral transduction of SZ95 cells with pBabepuro (d and g), pBabeΔNLef1 (e and h), and ΔNβ-catenin/T2 (f and i). Transduced gene products were detected by immunostaining with anti-Myc tag (e) and antihemagglutinin tag (d and f). Cells were labeled with antibodies specific for E-cadherin (to reveal cell-cell borders; red, g-i) and IHH (green, g-i). ΔNLef1 increased production of IHH protein (arrows in h). [Scale bars: 20 μm (b-i).] (j) Growth curve of cells transduced with pBabepuro (ev), pBabeΔNLef1 (dnLef1), or ΔNβ-catenin (T2). (k) Model of interactions between β-catenin and Hh signaling in epidermis. SHH promotes proliferation of progenitor cells of the hair lineage, whereas IHH stimulates proliferation of sebocyte precursors. The IHH signal is proposed to be produced by differentiated sebocytes and act on the sebocyte progenitors in a paracrine fashion.

β-Catenin Signaling Regulates Growth and IHH Expression of Human Sebocytes in Vitro. To investigate whether β-catenin signaling affected the growth and differentiation of sebocytes in vitro we infected SZ95 sebocytes with retroviral expression vectors for mutant proteins of the β-catenin signaling cascade. SZ95 sebocytes were infected with pBabeT2, a stabilized, N-terminally deleted β-catenin mutant (ΔNβ-catenin) (38), pBabeΔNLef1 to block β-catenin signaling (10) and the empty vector pBabepuro, as a control. Expression of the mutant proteins was analyzed by immunofluorescence staining with antibodies against the T2 hemagglutinin tag (Fig. 5 d and f) and the ΔNLef1 myc-tag (Fig. 5e), and by Western blotting (data not shown). As expected, ΔNLef1 could be readily detected in the nuclei of infected SZ95 cells (Fig. 5e), whereas ΔNβ-catenin (T2) was expressed in the cytoplasm, nucleus, and plasma membrane of infected cells (Fig. 5f). Cells expressing the empty vector (Fig. 5d) served as a negative control.

We analyzed whether manipulating β-catenin signaling in infected SZ95 sebocytes could regulate IHH synthesis. ΔNLef1 increased expression of IHH in SZ95 sebocytes compared with the empty vector control cells whereas ΔNβ-catenin slightly decreased expression of IHH (Figs. 4j Right and 5 g-i). Neither ΔNLef1 nor ΔNβ-catenin had any effect on sebocyte differentiation as evaluated by Nile red staining of low-density cultures (data not shown). ΔNLef1 stimulated proliferation of undifferentiated sebocytes compared with control cells (Fig. 5j). This finding is in contrast to the effects of the constructs on the growth of normal human IFE keratinocytes: ΔNLef1 decreases proliferation and ΔNβ-catenin stimulates proliferation of putative IFE stem cells in vitro (ref. 38 and data not shown). The stimulation of SZ95 cell proliferation by ΔNLef1 was abolished by cyclopamine (data not shown). Our results demonstrate that expression of ΔNLef1 increases IHH expression in differentiated sebocytes and stimulates proliferation of undifferentiated sebocytes.

Discussion

There is increasing evidence that Hh signaling can regulate growth and survival in both differentiated and undifferentiated cells (47-49) and can also regulate lineage-specific differentiation (50-52). Hh family members tend to act in a paracrine fashion (12).

We have shown that IHH is up-regulated in the differentiated sebocytes of normal human and mouse epidermis and in sebaceous tumors. The nuclear accumulation of GLI1 in undifferentiated sebocytes in vivo is an indication of active Hh signaling (53-56). The negative effect of cyclopamine on proliferation of undifferentiated sebocytes in vitro is direct evidence that IHH stimulates proliferation of sebaceous progenitor cells. Thus we propose that, whereas SHH promotes proliferation of progenitors of the hair lineages (20), IHH produced by mature sebocytes promotes proliferation of sebaceous progenitors in a paracrine manner (Fig. 5k). Whereas β-catenin levels regulate lineage choice within the epidermis (1), Hh promotes the proliferation of committed progenitors (Fig. 5k). SHH and IHH appear to signal via common receptors, and so specificity would be at the level of the cell type expressing each ligand (12).

We observed that ΔNLef1 increased IHH expression in cultured sebocytes, and we propose that this is one mechanism by which ΔNLef1 stimulates proliferation of sebocyte progenitors (Fig. 5k). The formation of spontaneous sebaceous tumors in K14ΔNLef1 mice is likely to be a consequence of the increased number of sebocyte progenitors, although whether, by analogy with the effects of overexpressing SHH (27-30), IHH overexpression is sufficient to induce the tumors remains to be tested. One mechanism by which SHH stimulates proliferation is by up-regulation of D-type cyclins (16, 57, 58). However, K14ΔNLef1 transgenic mouse tumors do not express cyclin D1, because β-catenin signaling is inhibited (10). An alternative mechanism by which IHH could drive sebaceous proliferation in the presence of ΔNLef1 would be by opposing Waf1/p21-mediated growth arrest (59).

Sebaceous tumors are a characteristic feature of Muir-Torre syndrome in which patients develop colon cancer (60). Muir-Torre patients and other individuals with colon cancer (sporadic and hereditary nonpolyposis colon cancer) have microsatellite instability. Interestingly, microsatellite instability in colon cancers can result in frameshift mutations in the TCF4 gene (61). These observations raise the intriguing possibility that sebaceous tumors of Muir-Torre patients might have frameshift mutations in the Lef1 gene.

Although IHH is linked primarily to cartilage differentiation (15) there are several reports that it plays an important role in epithelia. Furthermore, in both cartilage and epithelia there is an association between IHH and hormonal regulation. In mouse uterus, IHH mediates epithelial-mesenchymal interactions and is regulated by progesterone (62, 63). IHH is also expressed in mammary gland epithelium and is up-regulated during pregnancy and lactation (64). Our observation of IHH expression in sebocytes, which are under the control of androgens and a range of other hormones (65, 66), fits very well with these observations. It is also of interest that liganded androgen receptor represses β-catenin/T cell factor-mediated transcription (67-69), suggesting a possible synergy between ΔNLef1 and androgens in promoting sebocyte differentiation in addition to the potential synergy between androgens and IHH in stimulating proliferation of sebocyte progenitors.

Taken together, our data suggest that IHH and ΔNLef1 function in concert to change the proliferative and differentiation program of epidermal stem cell daughters. We are far from understanding the detailed interactions between Wnt and Hh signaling cascades in the epidermis and how they, in turn, interact with the numerous other signaling pathways that control epidermal stem cell fate (1, 2). The recent finding that in transgenic mouse epidermis the same β-catenin mutation exerts different effects depending on the cells in which it is expressed underlines the importance of cellular context and microenvironment in the control of tissue renewal and differentiation (11).