Original Paper

Oncogene (2004) 23, 1263–1274. doi:10.1038/sj.onc.1207240 Published online 22 December 2003

The zinc-finger transcription factor GLI2 antagonizes contact inhibition and differentiation of human epidermal cells

Gerhard Regl1, Maria Kasper1, Harald Schnidar1, Thomas Eichberger1, Graham W Neill2, Mohammed S Ikram2, Anthony G Quinn3, Mike P Philpott2, Anna-Maria Frischauf1 and Fritz Aberger1

  1. 1Institute of Genetics, University of Salzburg, Hellbrunner Strasse 34, A-5020 Salzburg, Austria
  2. 2Center for Cutaneaous Research, Barts and The London Queen Mary's School of Medicine & Dentistry, University of London, UK
  3. 3Experimental Medicine, AstraZeneca R & D Charnwood, Leicestershire, UK

Correspondence: F Aberger, E-mail: fritz.aberger@sbg.ac.at

Received 22 September 2003; Accepted 26 September 2003; Published online 22 December 2003.



In stratified epidermis, activation of the Hh/Gli signal transduction pathway has been implicated in the control of cell proliferation and tumorigenesis. The zinc-finger transcription factor Gli2 has been identified as critical mediator of the Hh signal at the distal end of the pathway, but the molecular mechanisms by which Gli2 regulates cell proliferation or induces epidermal malignancies such as basal cell carcinoma are still unclear. Here, we provide evidence for a role of human GLI2 in antagonizing contact inhibition and epidermal differentiation. We show by gene expression profiling that activation of the GLI2 oncogene in human keratinocytes activates the transcription of a number of genes involved in cell cycle progression such as E2F1, CCND1, CDC2 and CDC45L, while it represses genes associated with epidermal differentiation. Analysis of the proliferative effect of GLI2 revealed that GLI2 is able to induce G1–S phase progression in contact-inhibited keratinocytes. Detailed time-course experiments identified E2F1 as early transcriptional target of GLI2. Further, we show that GLI2 expression in human keratinocytes results in a marked downregulation of epidermal differentiation markers. The data suggest a role for GLI2 in Hh-induced epidermal neoplasia by opposing epithelial cell cycle arrest signals and epidermal differentiation.


hedgehog signalling, GLI proteins, keratinocytes, basal cell carcinoma, epidermal differentiation, cell cycle



The hedgehog (Hh)-signal transduction pathway, first identified by genetic analysis of Drosophila embryonic mutants, plays a critical role in a number of developmental processes, including pattern formation, control of cell differentiation, proliferation and growth (reviewed in Ingham and McMahon, 2001). The importance of Hh signalling in vertebrate embryonic development and in the control of cell proliferation is demonstrated by mutations in key components of the pathway, which account for severe congenital malformations and tumour development in man (reviewed in Goodrich and Scott, 1998; Toftgard, 2000; Mullor et al., 2002; Wetmore, 2003).

Basal cell carcinoma (BCC) of the skin represents one of the most common malignancies in the Western world. Ligand-independent activation of Hh signalling in epidermis by mutational inactivation of the Hh-receptor PTCH, which in the absence of ligand represses the pathway, or by activating mutations in the Hh-signal transducer SMOH has been implicated as the critical event in BCC development (Johnson et al., 1996; Stone et al., 1996; Xie et al., 1998). Further support for a causative role of inappropriate Hh signalling in BCC has come from studies of transgenic mice expressing Shh itself or a BCC-derived oncogenic form of SMOH in the basal layer of the epidermis. In either experiment, mice developed BCC-like structures, showing that activation of Hh-signal transduction in epidermal cells is sufficient to induce skin tumorigenesis (Oro et al., 1997; Xie et al., 1998).

The zinc-finger transcription factors (TFs) GLI1 and GLI2, which act at the very distal end of the Hh pathway, have been identified as putative mediators of Hh-induced neoplasia, since overexpression of either TF in the epidermis of transgenic mice induces various types of tumours, some of which show BCC-like features (Grachtchouk et al., 2000; Nilsson et al., 2000; Sheng et al., 2002). Their relative contribution to Hh-induced tumorigenesis is, however, unclear at present.

Although either TF is a potent oncogene in epidermal cells, evidence has accumulated suggesting that GLI2 rather than GLI1 may represent the primary mediator of the Hh signal during embryogenesis and tumorigenesis: firstly, Gli1 is dispensable for normal development and for Shh-induced medulloblastoma formation (Park et al., 2000; Weiner et al., 2002), while loss of Gli2 function results in severe developmental anomalies similar to those observed in Shh knockout mice (Ding et al., 1998; Matise et al., 1998; Mill et al., 2003). Secondly, removal of Gli2 but not of Gli1 can partially rescue the phenotype of patched knockout mice, which suffer from hyperactivation of the Hh pathway (Bai et al., 2002). Furthermore, Gli2 has been shown to act upstream of Gli1, since removal of Gli2 function decreases levels of Gli1 mRNA (Ding et al., 1998; Bai et al., 2002; Mill et al., 2003), and overexpression of GLI2 in epidermal cells results in induction of GLI1 expression (Regl et al., 2002).

While the genetic lesions involved in BCC development are well characterized, little is known about the downstream events leading to tumorigenic conversion of epidermal cells in response to inappropriate Hh signalling. Recent experiments addressing the mechanism by which Hh signalling controls cell proliferation have shown that the pathway can interact with the cell cycle machinery at various points. In Drosophila, Hh regulates cell proliferation in the developing eye by activating cyclin D and cyclin E expression. Analysis of the Drosophila cyclin E promoter showed that Cubitus interruptus – the Drosophila homologue of vertebrate GLI proteins – directly stimulates cyclin E transcription (Duman-Scheel et al., 2002). In vertebrates, Shh protein stimulates proliferation of cerebellar neuronal precursor cells by regulating the expression of D-type cyclins (Kenney and Rowitch, 2000), and in human keratinocytes expression of SHH has been shown to promote epidermal proliferation and oppose p21-induced epithelial cell cycle arrest (Fan and Khavari, 1999). Further, loss of Shh or Gli2 function in mice results in a significant decrease of proliferating cells in the hair follicle (Mill et al., 2003), and in vitro overexpression of GLI1 and GLI2 has been shown to stimulate S phase in human keratinocytes (Regl et al., 2002).

Although these data suggest that the proliferative effect of Shh and Gli proteins on neuronal and epidermal cells plays a critical role in brain and skin tumour development of vertebrates, the details of the molecular processes involved in Hh-mediated neoplasia remain to be established.

To elucidate the role of GLI2 in epidermal homeostasis and disease, we analysed the effect of GLI2 expression on the molecular phenotype of human keratinocytes. Using DNA-array technology, we show that GLI2 induces the expression of key regulators of cell cycle progression, while it represses genes associated with epidermal differentiation. Detailed analysis of the proliferation and differentiation properties of GLI2-expressing keratinocytes suggests a role of GLI2 in antagonizing both cell cycle arrest signals and epidermal differentiation. The data provide insight into the mechanism by which hyperactivation of GLI2 may lead to tumorigenic conversion of epidermal cells.



GLI2 induces regulators of cell cycle progression but represses differentiation-associated genes in the human keratinocyte cell line HaCaT

Although GLI2 has been implicated in Hh-induced BCC development, little is known about the mechanisms and target genes regulated by this transcription factor in human epidermis. Differences between murine and human skin and the relatively high resistance of mice to BCC development prompted us to use a purely human in vitro system to study the effect of GLI2 expression on epidermal cells in the absence of paracrine signals derived from dermal cells. To identify GLI2-regulated genes by DNA-array technology, we introduced into the human keratinocyte line HaCaT (Boukamp et al., 1988) a tetracycline-regulated GLI2 expression system (Figure 2d). This strategy ensures highly reproducible and temporally controlled transgene expression, which greatly facilitates data analysis of gene expression profiling experiments. The system also allows detailed time-course studies and conditional activation of GLI2 in quiescent cells, which is very difficult to achieve with retroviral expression systems.

Figure 2.
Figure 2 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

GLI2 fails to induce G1 exit in starved HaCaT cells. (a) tet-GLI2 HaCaT cells were starved overnight and subsequently cultured for 48 h either in the presence (48 h +tet) or absence (48 h -tet) of tetracycline or in the presence of 10% FBS (48 h +FBS) as positive control. The percentage of cells in G1 or S/G2 phase was analysed by flow cytometry. (b) Real-time RT–PCR analysis of mRNA levels of cell cycle regulators in GLI2-expressing HaCaT cells. Fold mRNA increase is expressed as the ratio of the mRNA level of a gene in tetracycline-treated to its mRNA level in untreated cells. (c) Western blot analysis of the Rb phosphorylation state in starved tet-inducible GLI2 HaCaT cells. After starving, cells were cultured for another 24 or 48 h either in the presence (24 h +tet; 48 h +tet) or absence (24 h -tet; 48 h -tet) of tetracycline. (d) Western blot analysis of tet-GLI2 HaCaT cells showing GLI2 protein expression in response to various times of tetracycline treatment (+)

Full figure and legend (100K)

We have chosen the spontaneously immortalized, nontumorigenic human keratinocyte line HaCaT as a model system, since, although aneuploid, it resembles primary human keratinocytes in that it has retained the capacity to differentiate and even to reconstitute stratified epidermis when grafted onto nude mice or used in organotypic cultures (Boukamp et al., 1988; Schoop et al., 1999). Further, we have shown that with respect to target gene expression, HaCaT and primary keratinocytes respond to GLI expression in a largely identical manner (Regl et al., 2002; data not shown). On the other hand, the presence of cytogenetic aberrations and the immortalized phenotype of HaCaT cells (Boukamp et al., 1988; Lehman et al., 1993) may limit the physiological relevance of data on cell cycle regulation. We, therefore, validated results obtained with HaCaT cells by using normal human keratinocytes expressing GLI2 via retroviral gene transfer.

To analyse the phenotypic changes of human keratinocytes in response to GLI2 expression on a molecular level, cDNA from confluent tetracycline-inducible GLI2-HaCaT cells (henceforth referred to as tet-GLI2 HaCaT), either treated with tetracycline for 96 h or untreated, was hybridized to high-density cDNA arrays containing a set of 2135 sequence-verified EST clones spotted in duplicate onto nylon membranes. Tetracycline-treated and -untreated samples were analysed on a total of four arrays, yielding eight data points for each gene. Results of all eight array evaluations were normalized for total signal intensity and used as input data for a two-class unpaired statistical analysis using significance analysis of microarrays (SAM) software (University of Stanford; see Materials and methods) (Tusher et al., 2001). This yielded a total of 107 differentially expressed genes (Table 2), with 36 genes induced (red label) and 71 genes repressed (green label) by GLI2. The results were comparable to data obtained with a pool of four independently isolated tet-GLI2-HaCaT lines, indicating that inter-clone variability is negligible (data not shown).

Expression of GLI2 led to an increase of mRNA levels of genes involved in cell cycle progression (Table 2, light-yellow background label) such as CDC2, E2F1, CCND1, PCNA and CDC45L, while mRNA levels of the CDK-inhibitor CDKNA1/p21 were reduced. In addition, genes known to be expressed either at elevated (PTCH, TNC, MMP2) or decreased (MYC, ITGA6, BPAG1, LAMB3) levels in BCC compared to normal skin (pink background label) (Savoia et al., 1993; Tuominen et al., 1997; Chopra et al., 1998; Varani et al., 2000; Bonifas et al., 2001; Regl et al., 2002) were appropriately regulated in tet-GLI2 HaCaT cells. We also found upregulation of interleukin-6 (IL6), a cytokine that has been shown to enhance the tumorigenicity of BCC (Jee et al., 2001).

To gain more insight into the biological processes regulated by GLI2 in human keratinocytes, we systematically screened various gene annotation and literature databases for expression data and possible functions of GLI2-regulated genes in epidermal cells (light-blue background). We found that a number of genes repressed by GLI2 had previously been shown to be expressed in differentiating or terminally differentiated keratinocytes of the cornified envelope (PI3 (also known as elavin), SPRRA2, CSTA, UGCG) (Zettergren et al., 1984; Steinert and Marekov, 1995; Takahashi et al., 1997; Watanabe et al., 1998) or to be involved in promoting keratinocyte differentiation including DLX3 (Morasso et al., 1996), the vitamin D receptor VDR (reviewed in Bikle et al., 2001), the Notch ligand JAG1 (Nickoloff et al., 2002) and the proto-oncogene JUNB (Welter et al., 1995; Rutberg et al., 1996; Medvedev et al., 1999). By contrast, GLI2 expression led to increased mRNA levels of integrin beta1 (ITGB1), a marker of undifferentiated keratinocytes in the basal layer of the epidermis (Jones et al., 1995).

Verification of array data by real-time RT–PCR analysis

The results obtained by DNA-array analysis were confirmed by real-time RT–PCR analysis of a selection of 18 genes either activated or repressed by GLI2. As shown in Figure 1, all 18 genes tested were found differentially expressed according to results from the array analysis. The higher fold-change values measured by real-time RT–PCR compared to the array approach is a common phenomenon and is likely to be due to lower background signals in the PCR-based approach.

Figure 1.
Figure 1 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Verification of DNA-array data by real-time RT–PCR analysis. Note that the fold change of mRNA levels in response to GLI2 has been plotted on a log 2 scale, due to high induction/repression levels of certain GLI2-regulated genes (e.g. PTCH (log 2)=4.4 means that PTCH mRNA levels increased 21.1-fold in response to GLI2 expression)

Full figure and legend (90K)

GLI2 fails to induce G1–S progression in the absence of growth factors

Having shown that GLI2 expression in human epidermal cells leads to a strong increase in transcript levels of key regulators of proliferation such as E2F1, CDC2 or CCND1, we next asked whether GLI2 has a stimulatory effect on cell cycle progression of keratinocytes independent of the presence of growth factors.

It has been well established that exposure of nondividing cells to mitogenic stimuli such as growth factors stimulates cell proliferation by increasing the levels of D-type cyclins, which can bind to and activate cyclin-dependent kinases such as CDK4 and CDK6. Active cyclin D/CDK complexes phosphorylate the retinoblastoma gene product Rb, which in its unphosphorylated state binds to and inhibits E2F proteins. Upon Rb phosphorylation, active E2F plays a central role in inducing the expression of S-phase genes and thus in promoting G1–S phase progression (for reviews see Harbour and Dean, 2000; Sherr, 2000).

To analyse whether GLI2 expression affects G1–Sphase progression in the absence of growth factors, tet-GLI2 HaCaT cells were first grown to 30% confluency, starved overnight in low-serum (0.2%) medium without tetracycline and then cultured for 48 h either in the presence or absence of tetracycline. As positive control, starved cells were grown for 48 h in medium complemented with 10% fetal bovine serum (FBS). The number of cells in G1 or S/G2 phase was subsequently determined by flow cytometry. As shown in Figure 2a, GLI2 expression did not lead to an increase in the number of cells in S/G2 phase compared to controls (34.4 and 33.8%, respectively), suggesting that in the absence of additional mitogenic signals GLI2 is unable to induce exit from Go/G1 arrest and re-entry into the cell cycle. By contrast, addition of 10% FBS to starved cells led to re-entry into the cell cycle as shown by the marked increase of cells in S/G2 phase (62.9%).

To investigate whether the inability of GLI2 to induce G1–S phase progression is due to a failure of GLI2 to activate cell cycle genes in starved cells, we analysed the changes in mRNA levels of cell cycle regulators in response to GLI2. To verify that GLI2 is appropriately activated and fully functional in starved cells, we analysed the mRNA induction of the known direct Gli-target gene PTCH as a control. PTCH transcription was induced by GLI2 to levels comparable to those observed in cells grown in complete medium (data not shown), indicating that GLI2 protein expression and activity is not altered by serum withdrawal. Figure 2b shows that treatment of starved tet-GLI2 HaCaT cells with tetracycline for 48 h does moderately increase the mRNA levels of genes involved in G1–S or S–M phase progression such as E2F1 (4.9-fold), CCNA2 (3.9-fold), CDC45L (6.1-fold), CCNB1 (2.7-fold), CKS1B (2.7-fold) and CDC2 (3.2-fold).

We also measured changes in the mRNA levels of D- and E-type cyclins, since at least some of these have been described as putative direct targets of Hh/Gli signalling (Duman-Scheel et al., 2002; Yoon et al., 2002) As shown in Figure 2b, expression of GLI2 in starved keratinocytes did not significantly change the mRNA levels of D-type (CCND1,2,3) and E-type cyclins (CCNE1,2,3: data not shown).

Since D-type cyclins are involved in the phosphorylation-mediated inactivation of Rb protein, the failure of GLI2 to induce G1–S progression in starved cells may be due to insufficient phosphorylation of Rb protein. To test this hypothesis, we analysed Rb phosphorylation using antibodies that distinguish between active (hypophosphorylated, pRb) and inactive (hyperphosphorylated, ppRb) Rb protein (Harbour and Dean, 2000). When starved cells were cultured for 24 or 48 h in the absence of tetracycline, only one distinct band corresponding to active Rb (pRb) was detected, indicating efficient cell cycle arrest. Activation of GLI2 expression for 24 or 48 h led to a moderate increase of inactive ppRb, but active pRb was still the predominant form, suggesting that Rb protein is not efficiently inactivated in starved GLI2-expressing cells (Figure 2c).

GLI2 induces re-entry into S phase in contact-inhibited human epidermal cells

Normal cells grown in vitro undergo cell cycle arrest as soon as they reach confluency, a process frequently suppressed in tumorigenic cells.

To analyse whether the GLI2 oncogene can antagonize contact inhibition of human keratinocytes, tet-GLI2 HaCaT cells were grown in complete media to confluency and cultured for another 36 h (t=0) without tetracycline to induce efficient contact inhibition. To test whether GLI2 is able to abrogate contact inhibition and induce re-entry into the cell cycle, contact-inhibited cells were cultured for another 48 h either in the presence or absence of tetracycline. Re-entry into the cell cycle was subsequently analysed by bromodeoxyuridine (BrdU) incorporation assays. Two independently isolated tet-GLI2 HaCaT lines (#28 and #31) were analysed to exclude clone-specific effects. As shown in Figure 3, only a small proportion of cells showed S-phase activity at 36 h post confluency (t=0 h) (Figure 3a and d (0 h), 9.2% of clone #28, 8.7% of clone #31). Confluent cultures grown for another 48 h in the absence of tetracycline showed a similar percentage of BrdU-positive cells (Figure 3b and d, #28: 8.1%; #31: 8.2%), while expression of GLI2 by tetracycline treatment (Figure 3c and d) caused a threefold increase of the number of cells in S phase (#28: 24.5%, #31: 24.5%), showing that GLI2 can antagonize contact inhibition and induce re-entry into the cell cycle.

Figure 3.
Figure 3 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

GLI2 induces reentry into G1–S phase in contact-inhibited keratinocytes. (ad): Analysis of cell cycle activity of contact-inhibited HaCaT cells in response to GLI2. BrdU incorporation was analysed as an indicator of G1–S phase progression. (a) Contact-inhibited tet-GLI2 HaCaT cells at 36 h post confluency (t=0 h) grown in the absence of tet (-tet). (bc) Contact-inhibited tet-GLI2 HaCaT cells grown for another 48 h either in the absence (b) or presence (c) of tetracycline. Note the marked increase of BrdU-positive cells in response to GLI2 expression (48 h, +tet). (d) Quantitative analysis of two independently isolated tet-GLI2 HaCaT lines (#28, #31) cultured as in (ac). Data represent the mean value calculated from triplicate experiments with 2000 cells counted in each replicate. (e) Real-time RT–PCR analysis of cell cycle regulators in contact-inhibited HaCaT cells expressing GLI2 for 6, 24 or 48 h. Note the rapid increase of E2F1 and decrease of CDKN1A mRNA levels. (f) Western blot analysis showing the Rb phosphorylation state in contact-inhibited tet-GLI2 HaCaT cells, grown for 24 or 48 h either in the presence or absence of tetracycline. Note that in GLI2-expressing cells (24 h +tet; 48 h +tet) only the inactive, hyperphosphorylated form of Rb (ppRb) is detected. (g) Real-time RT–PCR analysis of cell cycle regulators in contact-inhibited primary keratinocytes expressing GLI2. To induce contact inhibition, primary human keratinocytes were transduced with GLI2 or EGFP-control retrovirus, grown to confluency and cultured for another 24 h either in the absence (prim. KC) or presence of Ca2+ (prim. KC (Ca2+)). Fold mRNA induction values represent the ratio of mRNA levels in GLI2-expressing to mRNA levels in EGFP-expressing control cells

Full figure and legend (301K)

To relate the proliferative phenotype of GLI2-expressing keratinocytes to changes in gene expression and to identify early transcriptional changes in response to GLI2, we performed time-course studies of the mRNA levels of GLI2-regulated cell cycle genes. As shown in Figure 3e, mRNA levels of E2F1 were elevated threefold and those of the CDK-inhibitor p21/CDKN1A repressed fourfold already after 6 h of tetracycline treatment. Strong activation of other GLI2-regulated cell cycle genes was observed only after 24 and 48 h following tetracycline addition, suggesting that changes in E2F1 and CDKN1A mRNA levels represent early events in GLI2-induced G1–S progression. As in starved cells, mRNA levels of the suggested direct GLI target CCND2 (Yoon et al., 2002) were unchanged at all time points.

We then asked whether re-entry of contact-inhibited GLI2-expressing keratinocytes into the cell cycle is accompanied by efficient inactivation of Rb protein. Tet-GLI2 HaCaT cells were contact inhibited for 36 h and then grown for another 24 or 48 h either in the presence or absence of tetracycline. The phosphorylation state of Rb protein was subsequently analysed as described above. As shown in Figure 3f, only active hypophosphorylated pRb protein was present in contact-inhibited HaCaT cells grown in the absence of tetracycline, while expression of GLI2 led to efficient hyperphosphorylation and inactivation of Rb, respectively.

Next, we validated these results by analysing the expression of G1–S and G2–M phase progression genes in contact-inhibited primary human keratinocytes. Similar to HaCaT cells, expression of GLI2 in contact-inhibited primary keratinocytes (prim. KC) led to an increase of CCNA2, CDC2 and CCNB1 mRNA levels, while levels of CCND2 were largely unchanged (Figure 3g). Notably, treatment of GLI2-expressing contact-inhibited primary cells with 1 mM calcium (prim. KC (Ca2+)), a potent growth arrest and epidermal differentiation signal, was unable to abrogate the proliferative activity of GLI2. Under these conditions, the effect of GLI2 on growth-arrested keratinocytes was even enhanced compared to untreated contact-inhibited cells (prim. KC) as can be seen by the stronger increase of E2F2, CCND1 and CCNB1 mRNA levels. Again CCND2 mRNA levels were only very weakly induced by GLI2. Together, these results suggest that GLI2 opposes contact inhibition of epidermal cells by promoting G1–S and G2–M phase progression.

GLI2 represses expression of epidermal differentiation genes

The results of gene expression profiling revealed that a considerable number of genes repressed by GLI2 are normally expressed in differentiating or differentiated keratinocytes. This suggested that GLI2 may be able to suppress epidermal differentiation. To analyse whether GLI2 antagonizes keratinocyte differentiation, tet-GLI2 HaCaT cells were grown to confluency in high-calcium medium (1.2 mM Ca2+) – a condition that increases the expression of epidermal differentiation markers (Garach-Jehoshua et al., 1998) – and then treated with tetracycline to induce GLI2 expression for 6, 24 or 48 h. As indicator of cell differentiation, the mRNA levels of early (Keratin 1 and 10 (KRT1 and KRT10)) and late epidermal differentiation markers (Involucrin (IVL) and small proline rich protein 2A (SPRR2A)) and of the basal cell/stem cell marker beta 1 integrin (ITGB1) were measured by real-time RT–PCR and compared to uninduced control cells. As shown in Figure 4a, expression of GLI2 dramatically reduced the mRNA levels of all differentiation markers analysed. In contrast, the amount of ITGB1 mRNA increased in response to GLI2 (2.6-fold after 24 h and 3.6-fold after 48 h of tetracycline treatment). The repressive activity of GLI2 on the differentiation of epidermal cells was confirmed by retroviral expression of GLI2 in primary human keratinocytes. As in tet-GLI2 HaCaT cells, expression of GLI2 in primary human keratinocytes led to downregulation of mRNA levels of KRT10 (10.5-fold repressed), SPRR2A (3.6-fold repressed), the terminal differentiation marker Loricrin (LOR) (2.9-fold repressed) and IVL (2.5-fold repressed) (Figure 4b). The weaker repressive effect of GLI2 observed in primary cells compared to HaCaT cells is likely to be due to the lower GLI2 expression level achieved by the retroviral expression system (data not shown).

Figure 4.
Figure 4 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

GLI2 represses epidermal marker gene expression in HaCaT and primary human keratinocytes. (a) Real-time RT–PCR analysis of epidermal differentiation markers and the basal cell marker ITGB1 in response to GLI2 expression in tet-inducible HaCaT cells. (b) mRNA levels of epidermal differentiation markers in primary human keratinocytes transduced with a GLI2-expressing retrovirus (SIN-GLI2-EGFP). As reference, primary keratinocytes were transduced with EGFP-expressing retrovirus

Full figure and legend (111K)



Control of epidermal gene expression by the human oncogene GLI2

The development and maintenance of stratified epidermis requires a precise balance between cell proliferation and differentiation, which is controlled by reciprocal paracrine signalling events involving epidermal and underlying dermal cells (Werner and Smola, 2001). The complexity of these signalling processes makes it difficult to study cell-autonomous effects of a given gene in vivo. Using a purely human model system to analyse on a molecular level the phenotypic changes of keratinocytes in response to GLI2 expression in the absence of any dermal signals, we have shown that GLI2 activates genes involved in cell cycle progression, while it represses differentiation-associated genes. Further, expression of molecular BCC markers such as PTCH, TNC, MYC or BPAG1 was appropriately regulated in our experimental system. These results demonstrate the relevance of our in vitro model system to Hh/Gli-induced gene expression patterns in BCC.

Regulation of G1–S phase progression by GLI2

Taking advantage of our inducible GLI2 expression system, we have shown that GLI2 promotes G1–S progression in contact-inhibited keratinocytes. Notably, GLI2 does not simply delay the onset of contact inhibition but induces quiescent cells to re-enter the cell cycle. Detailed time-course studies of GLI2 target gene expression in contact-inhibited keratinocytes suggest that transcriptional activation of E2F1 and repression of CDKN1A/p21 are early responses to GLI2 expression. Compared to E2F1, whose overexpression has been shown to induce G1–S progression and DNA replication (Arata et al., 2000), the induction of CCNA2, CDC2, CDC45L and CCNE transcription was clearly delayed. Since all of these genes have been shown to be direct targets of E2F proteins (Ohtani et al., 1995; Helin, 1998; Arata et al., 2000; Muller et al., 2001; Ren et al., 2002), their induction in GLI2-expressing cells is likely to be the consequence of increased levels of active E2F protein that has been relieved from inhibition by Rb.

Analysis of homozygous mutant Gli2 mice has revealed a role of murine Gli2 in the activation of D-type cyclin expression in epidermal cells (Mill et al., 2003). A possible direct role of Gli proteins in regulating D-type cyclins has also been suggested by the presence of a putative Gli-binding site in the human CCND2 promoter that can be bound by recombinant GLI protein in vitro (Yoon et al., 2002). In our studies, however, expression of GLI2 in cultured human keratinocytes did not lead to significantly elevated mRNA levels of CCND2. Whether the inability of GLI2 to induce CCND2 expression is due to the in vitro system used or due to cell type-specific effects remains to be addressed in further studies.

In contrast to its proliferative effect in contact-inhibited keratinocytes, GLI2 failed to induce re-entry into the cell cycle in starved keratinocytes. This is likely to be due to the failure of GLI2 to cause sufficient inhibition of Rb function, which would be required for E2F1 protein to become fully active. Indeed, activation of E2F target genes in starved keratinocytes is much weaker compared to contact-inhibited cells, where Rb protein is inhibited efficiently by GLI2 expression (Harbour and Dean, 2000; Sherr, 2000). Our results obtained with starved GLI2-expressing human keratinocytes are consistent with studies of the mitogenic effect of Shh protein on cerebellar granular precursor cells (CGPC). Although treatment of CGPC with Shh induces proliferation and D-type cyclin expression in the presence of growth factors, it fails to induce G1–S progression in starved CGPC, suggesting that additional mitogenic signals are required for Hh/Gli-induced proliferation (Kenney and Rowitch, 2000). In the epidermis, such a signal may be required for activation of the PDGF-receptor alpha (PDGFRA), since inhibition of PDGFRA signalling in a Ptc -/- murine BCC cell line significantly reduces cell proliferation (Xie et al., 2001).

GLI2 represses epidermal differentiation

In addition to the proliferative effect on contact-inhibited keratinocytes, we have uncovered a role of GLI2 in opposing differentiation of human keratinocytes, as evidenced by the strong repression of epidermal differentiation markers in GLI2-expressing HaCaT and primary human keratinocytes. Although these data are based on in vitro expression systems, it is noteworthy that in vivo human GLI2 is predominantly expressed in regions of normal skin and BCC, where undifferentiated and/or (hyper)proliferating epidermal cells are located (Ikram et al., submitted).

The strong repressive effect of GLI2 on the expression of a number of genes – an effect that is not observed by expression of GLI1 (Eichberger et al., in preparation) – points to the existence of a GLI2 repressor form which, like the Drosophila Gli homologue Cubitus interruptus, may be generated by proteolytic processing of GLI2 full-length protein (Aza-Blanc et al., 1997). Murine Gli2 has been shown to contain a putative N-terminal repressor domain, since removal of this domain converts Gli2 from a weak into a strong transcriptional activator (Sasaki et al., 1999). In contrast, however, human GLI2 lacks this domain, which may explain the strong activator function of full-length GLI2 protein (Regl et al., 2002) and also point to different modes of Gli2 processing in murine and human epidermis.

Taken together, the results presented here suggest a model, where GLI2 plays a dual role as activator of keratinocyte proliferation and repressor of epidermal differentiation. In normal skin, both functions may be implicated in the maintenance of epidermal homeostasis, while in state of disease, hyperactivation of GLI2 function by inappropriate Hh signalling may distort this balance leading to epidermal neoplasia. A detailed understanding of the oncogenic processes governed by GLI2 will be a major future aim and involve the identification and functional analysis of direct target genes in human epidermis.


Materials and methods

Cell culture and retroviral infection of keratinocytes

The T-REx system (Invitrogen) was used to generate tet-GLI2 HaCaT lines expressing N-terminally His-tagged GLI2 under the control of the tetracycline repressor. GLI2 expression was induced by adding 1 mg/l tetracycline (Invitrogen) to the cell culture medium (high-glucose Dulbecco's modified Eagle's medium (Life Technologies) supplemented with 10% FBS, 100 mg/l streptomycin and 62.5 mg/l penicillin). Real-time RT–PCR analysis of independently isolated tet-GLI2 HaCaT lines showed that addition of tetracycline increased GLI2 mRNA to levels comparable to those in BCC samples (data not shown), suggesting that the inducible system is suitable to simulate in vitro GLI gene expression in diseased skin. Further, it indicates that the system does not produce unphysiological transgene expression levels, which otherwise may result in unspecific results due to cytotoxicity or other stress responses.

To generate the retroviral bicistronic GLI2-EGFP expression construct, N-terminally His-tagged GLI2 was cloned into pI2E-A, a modified version of the pIRES2-EGFP plasmid (Clontech) (Regl et al., 2002). PI2E-A-GLI2 was digested with SalI and NotI to excise CMV-GLI2-IRES-EGFP. The resulting fragment was cloned into XhoI–NotI-digested retroviral SIN-IP plasmid (gift from Prof. P Khavari) to create SIN-GLI2-EGFP. Amphotropic retrovirus was produced as previously described (Deng et al., 1997) except that the more efficient Phoenix packaging cell line was used. Primary human keratinocytes were isolated and infected as previously described (Rheinwald and Green, 1975; Deng et al., 1997) except that cells were plated at a density of 0.5–1.0 times 106 cells per 10 cm dish in defined keratinocyte-SFM (Invitrogen) 16–18 h prior to infection.

RNA isolation and real-time RT–PCR analysis

Total RNA of HaCaT cells and primary human keratinocytes was isolated using TRI reagent (Molecular Resarch Center) followed by LiCl precipitation. RNA used for real-time RT–PCR analysis was further purified with the High Pure RNA Isolation Kit (Roche) to remove any genomic contaminations. cDNA was synthesized with Superscript II (Rnase H-) reverse transcriptase (Invitrogen), according to the manufacturer's instructions. Real-time RT–PCR analysis was performed on a Rotorgene 2000 (Corbett Research) using iQTM SYBR Green Supermix (BIORAD). Real-time primer sequences are shown in Table 1 or have been published previously (Regl et al., 2002). Absence of genomic DNA was ensured by omitting reverse transcriptase during cDNA synthesis. The specificity and quality of the PCR reactions was controlled by direct sequencing of the PCR amplicons, melting curve analysis and gel electrophoresis. 'Primer-only' controls were done to ensure the absence of primer dimers. Large ribosomal protein P0 (RPLP0) was used as a reference standard for all analyses to control for the amount of sample material (Martin et al., 2001). Real-time analysis using cyclophilin E as reference standard gave comparable results (data not shown).

Gene expression profiling

A nonredundant set of 2135 sequence-verified EST clones selected from the human UniGem V2.0 library (Incyte Genomics Inc.) and the RZPD Unigene clone collection (http://www.rzpd.de) was PCR-amplified and spotted in duplicate on nylon membranes (Hybond N+, Amersham Biosciences) using a MicroGridII spotting robot (BioRobotics, UK). cDNA labelling and array hybidizations were carried out as described previously (Aberger et al., 2001). In brief, 15 mug total RNA from each sample was reverse transcribed with Superscript II (Rnase H-) reverse transcriptase (Invitrogen) in the presence of 70 muCi alpha-[33P]dCTP (3000 Ci/mmol, Amersham Biosciences). Labelled cDNA was purified with GFX DNA purification columns (Amersham Biosciences). Arrays were hybridized for 36 h at 68°C in 10 ml hybridization buffer (5 times SSC, 5 times Denhard's, 1% SDS) and washed for 20 min at 60°C once in 2 times SSC/0.1% SDS, and twice in 0.2 times SSC/0.1% SDS. Filters were exposed for 4 days and scanned with a BAS-1800II (Fuji) phosphoimager. Images were analysed using the AIDA Metrix suite (Raytest). Data were normalized for total signal intensity and statisically analysed using SAM software (University of Stanford; Tusher et al., 2001). For SAM analysis, the delta value was set to represent a minimal false detection rate (FDR) of 0.17%. The stringency was further increased by setting the fold-change parameter to 1.8, such that only those genes were called significant that showed at least a 1.8-fold induction or repression in response to GLI2 expression. Furthermore, only if both duplicates were called significant by SAM, the gene was considered differentially expressed and included in Table 2.

BrdU incorporation assays and flow cytometry analysis

BrdU incorporation assays were performed with the FLUOS in situ cell proliferation kit (Roche). Cells were incubated for 90 min in the presence of BrdU before detection with flourescein-labelled anti-BrdU antibody. Microscopic imaging was carried out on an Olympus IX 70 microscope equipped with a SPOT CCD-camera (Diagnostic Instruments Inc.). Flow cytometry analysis was performed on a FACSCalibur (Becton Dickenson) and data were analysed with CellQuest software (Becton Dickenson).

Western blot analysis

N-terminally His-tagged GLI2 protein was detected with a monoclonal peroxidase HRP-conjugated anti-polyhistidine antibody (HIS-1, Sigma-Aldrich). Hypo- and hyperphosphorylated Rb protein (pRb and ppRb) was detected using a monoclonal mouse anti-human Rb antibody (G3-245, BD Biosciences) and a secondary sheep anti-mouse HRP-conjugated antibody (Amersham Biosciences). Proteins were visualized with ECL detection system (Amersham Biosciences).



  1. Aberger F, Costa-Pereira AP, Schlaak JF, Williams TM, O'Shaughnessy RF, Hollaus G, Kerr IM and Frischauf AM. (2001). Genomics, 77, 50–57. | Article | PubMed | ISI | ChemPort |
  2. Arata Y, Fujita M, Ohtani K, Kijima S and Kato JY. (2000). J. Biol. Chem., 275, 6337–6345. | Article | PubMed | ChemPort |
  3. Aza-Blanc P, Ramirez-Weber FA, Laget MP, Schwartz C and Kornberg TB. (1997). Cell, 89, 1043–1053. | Article | PubMed | ChemPort |
  4. Bai CB, Auerbach W, Lee JS, Stephen D and Joyner AL. (2002). Development, 129, 4753–4761. | Article | PubMed | ISI | ChemPort |
  5. Bikle DD, Ng D, Tu CL, Oda Y and Xie Z. (2001). Mol. Cell. Endocrinol., 177, 161–171. | Article | PubMed | ISI | ChemPort |
  6. Bonifas JM, Pennypacker S, Chuang PT, McMahon AP, Williams M, Rosenthal A, De Sauvage FJ and Epstein Jr EH. (2001). J. Invest. Dermatol., 116, 739–742. | Article | PubMed | ISI | ChemPort |
  7. Boukamp P, Petrussevska RT, Breitkreutz D, Hornung J, Markham A and Fusenig NE. (1988). J. Cell Biol., 106, 761–771. | Article | PubMed | ISI | ChemPort |
  8. Chopra A, Maitra B and Korman NJ. (1998). J. Invest. Dermatol., 110, 52–56. | Article | PubMed | ISI | ChemPort |
  9. Deng H, Lin Q and Khavari PA. (1997). Nat. Biotechnol., 15, 1388–1391. | Article | PubMed | ISI | ChemPort |
  10. Ding Q, Motoyama J, Gasca S, Mo R, Sasaki H, Rossant J and Hui CC. (1998). Development, 125, 2533–2543. | PubMed | ISI | ChemPort |
  11. Duman-Scheel M, Weng L, Xin S and Du W. (2002). Nature, 417, 299–304. | Article | PubMed | ISI | ChemPort |
  12. Fan H and Khavari PA. (1999). J. Cell Biol., 147, 71–76. | Article | PubMed | ISI | ChemPort |
  13. Garach-Jehoshua O, Ravid A, Liberman UA, Reichrath J, Glaser T and Koren R. (1998). Br. J. Dermatol., 139, 950–957. | Article | PubMed | ISI | ChemPort |
  14. Goodrich LV and Scott MP. (1998). Neuron, 21, 1243–1257. | Article | PubMed | ISI | ChemPort |
  15. Grachtchouk M, Mo R, Yu S, Zhang X, Sasaki H, Hui CC and Dlugosz AA. (2000). Nat. Genet., 24, 216–217. | Article | PubMed | ISI | ChemPort |
  16. Harbour JW and Dean DC. (2000). Genes Dev., 14, 2393–2409. | Article | PubMed | ISI | ChemPort |
  17. Helin K. (1998). Curr. Opin. Genet. Dev., 8, 28–35. | Article | PubMed | ISI | ChemPort |
  18. Ingham PW and McMahon AP. (2001). Genes Dev., 15, 3059–3087. | Article | PubMed | ISI | ChemPort |
  19. Jee SH, Shen SC, Chiu HC, Tsai WL and Kuo ML. (2001). Oncogene, 20, 198–208. | Article | PubMed | ISI | ChemPort |
  20. Johnson RL, Rothman AL, Xie J, Goodrich LV, Bare JW, Bonifas JM, Quinn AG, Myers RM, Cox DR, Epstein Jr EH and Scott MP. (1996). Science, 272, 1668–1671. | Article | PubMed | ISI | ChemPort |
  21. Jones PH, Harper S and Watt FM. (1995). Cell, 80, 83–93. | Article | PubMed | ISI | ChemPort |
  22. Kenney AM and Rowitch DH. (2000). Mol. Cell. Biol., 20, 9055–9067. | Article | PubMed | ISI | ChemPort |
  23. Lehman TA, Modali R, Boukamp P, Stanek J, Bennett WP, Welsh JA, Metcalf RA, Stampfer MR, Fusenig N and Rogan EM et al. (1993). Carcinogenesis, 14, 833–839. | Article | PubMed | ISI | ChemPort |
  24. Martin KJ, Graner E, Li Y, Price LM, Kritzman BM, Fournier MV, Rhei E and Pardee AB. (2001). Proc. Natl. Acad. Sci. USA, 98, 2646–2651. | Article | PubMed | ChemPort |
  25. Matise MP, Epstein DJ, Park HL, Platt KA and Joyner AL. (1998). Development, 125, 2759–2770. | PubMed | ISI | ChemPort |
  26. Medvedev A, Saunders NA, Matsuura H, Chistokhina A and Jetten AM. (1999). J. Biol. Chem., 274, 3887–3896. | Article | PubMed | ISI | ChemPort |
  27. Mill P, Mo R, Fu H, Grachtchouk M, Kim PC, Dlugosz AA and Hui CC. (2003). Genes Dev., 17, 282–294. | Article | PubMed | ISI | ChemPort |
  28. Morasso MI, Markova NG and Sargent TD. (1996). J. Cell Biol., 135, 1879–1887. | Article | PubMed | ISI | ChemPort |
  29. Muller H, Bracken AP, Vernell R, Moroni MC, Christians F, Grassilli E, Prosperini E, Vigo E, Oliner JD and Helin K. (2001). Genes Dev., 15, 267–285. | Article | PubMed | ISI | ChemPort |
  30. Mullor JL, Sanchez P and Altaba AR. (2002). Trends Cell Biol., 12, 562–569. | Article | PubMed | ISI | ChemPort |
  31. Nickoloff BJ, Qin JZ, Chaturvedi V, Denning MF, Bonish B and Miele L. (2002). Cell Death Differ., 9, 842–855. | Article | PubMed | ISI | ChemPort |
  32. Nilsson M, Unden AB, Krause D, Malmqwist U, Raza K, Zaphiropoulos PG and Toftgard R. (2000). Proc. Natl. Acad. Sci. USA, 97, 3438–3443. | Article | PubMed | ChemPort |
  33. Ohtani K, DeGregori J and Nevins JR. (1995). Proc. Natl. Acad. Sci. USA, 92, 12146–12150. | Article | PubMed | ChemPort |
  34. Oro AE, Higgins KM, Hu Z, Bonifas JM, Epstein Jr EH and Scott MP. (1997). Science, 276, 817–821. | Article | PubMed | ISI | ChemPort |
  35. Park HL, Bai C, Platt KA, Matise MP, Beeghly A, Hui CC, Nakashima M and Joyner AL. (2000). Development, 127, 1593–1605. | PubMed | ISI | ChemPort |
  36. Regl G, Neill GW, Eichberger T, Kasper M, Ikram MS, Koller J, Hintner H, Quinn AG, Frischauf AM and Aberger F. (2002). Oncogene, 21, 5529–5539. | Article | PubMed | ISI | ChemPort |
  37. Ren B, Cam H, Takahashi Y, Volkert T, Terragni J, Young RA and Dynlacht BD. (2002). Genes Dev., 16, 245–256. | Article | PubMed | ISI | ChemPort |
  38. Rheinwald JG and Green H. (1975). Cell, 6, 331–343. | Article | PubMed | ISI | ChemPort |
  39. Rutberg SE, Saez E, Glick A, Dlugosz AA, Spiegelman BM and Yuspa SH. (1996). Oncogene, 13, 167–176. | PubMed | ISI | ChemPort |
  40. Sasaki H, Nishizaki Y, Hui C, Nakafuku M and Kondoh H. (1999). Development, 126, 3915–3924. | PubMed | ISI | ChemPort |
  41. Savoia P, Trusolino L, Pepino E, Cremona O and Marchisio PC. (1993). J. Invest. Dermatol., 101, 352–358. | Article | PubMed | ISI | ChemPort |
  42. Schoop VM, Mirancea N and Fusenig NE. (1999). J. Invest. Dermatol., 112, 343–353. | Article | PubMed | ISI | ChemPort |
  43. Sheng H, Goich S, Wang A, Grachtchouk M, Lowe L, Mo R, Lin K, de Sauvage FJ, Sasaki H, Hui CC and Dlugosz AA. (2002). Cancer Res., 62, 5308–5316. | PubMed | ISI | ChemPort |
  44. Sherr CJ. (2000). Cancer Res., 60, 3689–3695. | PubMed | ISI | ChemPort |
  45. Steinert PM and Marekov LN. (1995). J. Biol. Chem., 270, 17702–17711. | Article | PubMed | ISI | ChemPort |
  46. Stone DM, Hynes M, Armanini M, Swanson TA, Gu Q, Johnson RL, Scott MP, Pennica D, Goddard A, Phillips H, Noll M, Hooper JE, de Sauvage F and Rosenthal A. (1996). Nature, 384, 129–134. | Article | PubMed | ISI | ChemPort |
  47. Takahashi H, Kinouchi M, Wuepper KD and Iizuka H. (1997). J. Invest. Dermatol., 108, 843–847. | PubMed |
  48. Toftgard R. (2000). Cell. Mol. Life Sci., 57, 1720–1731. | Article | PubMed | ISI | ChemPort |
  49. Tuominen H, Pollanen R and Kallioinen M. (1997). J. Cutan. Pathol., 24, 590–596. | PubMed |
  50. Tusher VG, Tibshirani R and Chu G. (2001). Proc. Natl. Acad. Sci. USA, 98, 5116–5121. | Article | PubMed | ChemPort |
  51. Varani J, Hattori Y, Chi Y, Schmidt T, Perone P, Zeigler ME, Fader DJ and Johnson TM. (2000). Br. J. Cancer, 82, 657–665. | Article | PubMed | ISI | ChemPort |
  52. Watanabe R, Wu K, Paul P, Marks DL, Kobayashi T, Pittelkow MR and Pagano RE. (1998). J. Biol. Chem., 273, 9651–9655. | Article | PubMed | ISI | ChemPort |
  53. Weiner HL, Bakst R, Hurlbert MS, Ruggiero J, Ahn E, Lee WS, Stephen D, Zagzag D, Joyner AL and Turnbull DH. (2002). Cancer Res., 62, 6385–6389. | PubMed | ISI | ChemPort |
  54. Welter JF, Crish JF, Agarwal C and Eckert RL. (1995). J. Biol. Chem., 270, 12614–12622. | Article | PubMed | ISI | ChemPort |
  55. Werner S and Smola H. (2001). Trends Cell Biol., 11, 143–146. | Article | PubMed | ISI | ChemPort |
  56. Wetmore C. (2003). Curr. Opin. Genet. Dev., 13, 34–42. | Article | PubMed | ISI | ChemPort |
  57. Xie J, Aszterbaum M, Zhang X, Bonifas JM, Zachary C, Epstein E and McCormick F. (2001). Proc. Natl. Acad. Sci. USA, 98, 9255–9259. | Article | PubMed | ChemPort |
  58. Xie J, Murone M, Luoh SM, Ryan A, Gu Q, Zhang C, Bonifas JM, Lam CW, Hynes M, Goddard A, Rosenthal A, Epstein Jr EH and de Sauvage FJ. (1998). Nature, 391, 90–92. | Article | PubMed | ISI | ChemPort |
  59. Yoon JW, Kita Y, Frank DJ, Majewski RR, Konicek BA, Nobrega MA, Jacob H, Walterhouse D and Iannaccone P. (2002). J. Biol. Chem., 277, 5548–5555. | Article | PubMed | ISI | ChemPort |
  60. Zettergren JG, Peterson LL and Wuepper KD. (1984). Proc. Natl. Acad. Sci. USA, 81, 238–242. | Article | PubMed | ChemPort |


We are grateful to Sabine Siller for excellent technical assistance, to Drs Harald Esterbauer and Alexandra Kaser for critical reading of the manuscript and to Prof. Helmut Hintner for supply with tumour and normal skin material. This work was supported by FWF project P14227 (Austria), the University of Salzburg Schwerpunkt 'Biomedizin und Gesundheit', the Stiftungs- und Foerderungsgesellschaft of the University of Salzburg and by an EMBO short-term fellowship to TE.