Cooperation Between GLI and JUN Enhances Transcription of JUN and Selected GLI Target Genes

Home , GLI1, GLI2, PTCH1

Oncogene (2009) 28, 1639–1651 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE Cooperation between GLI and JUN enhances transcription of JUN and selected GLI target genes

S Laner-Plamberger1, A Kaser1, M Paulischta1, C Hauser-Kronberger2, T Eichberger1 and AM Frischauf1

1Department of Molecular Biology, University of Salzburg, Salzburg, Austria and 2Department of Pathology, St Johann’s Hospital, Paracelsus Medical University Salzburg, Salzburg, Austria

Sustained Hedgehog (HH) signaling is implicated in basal nents in regulating proliferation and differentiation is cell carcinoma of the skin and other types of cancer. Here reflected in the growing number ofmalignancies ofskin, we show that GLI1 and GLI2, the main transcriptional brain, pancreas, prostate and lung, which have been activators of the HH pathway, directly regulate expre- linked to ectopic expression ofactivator GLI proteins ssion of the activator protein 1 (AP-1) family member (reviewed in Evangelista et al., 2006; Hebrok, 2003; JUN, a transcription factor controlling keratinocyte Pasca di Magliano and Hebrok, 2003; Rubin and de proliferation and skin homeostasis. Activation of the Sauvage, 2006; Ruiz i Altaba et al., 2007; Stecca and JUN promoter by GLI is dependent on a GLI-binding site Ruiz i Altaba, 2002; Vezina and Bushman, 2007). and the AP-1 sites known to be involved in self-activation Normally, binding ofHH to its receptor PTCH leads of JUN. Transcription of JUN is greatly enhanced in the to the stabilization and activation offull-length GLI presence of GLI and requires activated JUN protein. proteins (GLIact) through the transmembrane signal GLI2act is a more potent activator than GLI1 in these transducer smoothened (SMOH). Activated GLI is, experiments and physical interaction with phosphorylated however, found not only in the presence of ligand, but JUN was only detected for GLI2act. The synergistic also when the pathway is activated by oncogenic events: effect of GLI and JUN extends to the activation of further inactivating mutations in PTCH (Hahn et al., 1996; GLI target genes as shown by shRNA-mediated knock- Johnson et al., 1996) and activating mutations in SMOH down of JUN in human keratinocytes. Some of these cause basal cell carcinoma (BCC) ofthe skin (Xie et al., cooperatively activated genes are involved in cell-cycle 1998), and mouse models ofBCC have been created progression, which is consistent with a significant reduction using transgenic mice expressing GLI1 or GLI2 in basal of the proliferative potential of GLI in the absence of JUN. keratinocytes ofthe skin (Grachtchouk et al., 2000; These results suggest a novel connection between HH/GLI Nilsson et al., 2000; Sheng et al., 2002). pathway activity and JUN, which may contribute to the In the absence ofHH, the GLI proteins are oncogenic activity of HH/GLI signaling in skin. inactivated and/or processed into repressor forms Oncogene (2009) 28, 1639–1651; doi:10.1038/onc.2009.10; (GLIrep). Although all GLI proteins bind the same published online 16 February 2009 consensus sequence with their highly conserved zinc- finger DNA-binding domain (Kinzler and Vogelstein, Keywords: GLI1; GLI2; Hedgehog signaling; activated 1990; Hallikas et al., 2006), the functional consequences JUN; basal cell carcinoma; keratinocytes may differ. GLI1 and GLI2 mostly act as activators of transcription whereas proteolytically processed GLI3 mainly acts as repressor. In contrast to GLI2 and GLI3, GLI1 is not directly activated by HH signaling but is Introduction itselfa transcriptional target gene ofGLI2 and GLI3. A wealth ofdata has shown that stability, cellular The GLI transcription factors are the main mediators localization and transcriptional activity ofGLI proteins ofthe Hedgehog (HH) signal and are important are regulated by several phosphorylation events and by during development, tissue homeostasis, regeneration other interacting proteins (reviewed in Hooper and and stem cell maintenance (reviewed in Beachy et al., Scott, 2005; Huangfu and Anderson, 2006; Ingham and 2004; Ingham and McMahon, 2001; Pasca di Magliano McMahon, 2001; Kasper et al., 2006a). and Hebrok, 2003; Ruiz i Altaba et al., 2007). The Recently, it has been shown that GLI activity is also importance ofthe HH/GLI pathway and its compo- responsive to other signaling pathways. GLI target gene activation and specificity was shown to be affected by Correspondence: Dr A-M Frischauf, Department of Molecular PKCd (Riobo et al., 2006a; Lauth et al., 2007), Akt/ Biology, University ofSalzburg, Hellbrunnerstr 34, Salzburg A-5020, PI3K (Riobo et al., 2006b), RAS/MEK/ERK (Pasca Austria. di Magliano et al., 2006; Stecca et al., 2007; Neill E-mail: [email protected] Received 4 August 2008; revised 11 December 2008; accepted 31 et al., 2008) and EGF signaling (Bigelow et al., 2005; December 2008; published online 16 February 2009 Kasper et al., 2006b; Mimeault et al., 2007). However, Cooperative target gene regulation by GLI and JUN S Laner-Plamberger et al 1640 the integration of different upstream signals by keratinocytes using HaCaT cell lines expressing either co-activators and other modulators on GLI target gene GLI1 (GLI1HaCaT) (Eichberger et al., 2006) or an promoters is still far from clear. activator form of GLI2 (GLI2act) (GLI2actHaCaT) (Regl JUN belongs to a protein family that forms homo- et al., 2004a) under doxycycline control. qRT–PCR dimers or heterodimers with members ofthe FOS, the showed that JUN mRNA is already elevated 12 h (four- activating transcription factor (ATF) and the musculoa- fold) after GLI2act induction and increases strongly at poneurotic fibrosarcoma families, constituting the acti- later time points (up to 16-fold at 72 h for GLI2act) vator protein 1 (AP-1) complex (reviewed in Hess et al., (Figure 1a). In HaCaT cells expressing GLI1, JUN 2004; Mechta-Grigoriou et al., 2001). JUN acts primarily mRNA levels are only moderately increased compared as activator oftranscription and is regulated at different to GLI2act (Figure 1a), even though levels ofthe GLI levels including activation ofits own transcription thus target PTCH (Figure 1b) are comparable. The same is true creating a positive feedback loop (Angel et al., 1988a). for GLI1 and GLI2act protein levels (Figures 1c and d, Phosphorylation by JUN N-terminal kinase (JNK), a upper panels). We found no significant change in mRNA target ofmitogen-activated protein kinase signaling, is levels ofthe AP-1 components JUNB and JUND and crucial for full transcriptional activity of JUN (reviewed FOS, FOSB, FRA1 and FRA2 (Supplementary Figure in Weston and Davis, 2007). Activated JUN mediates the S1). Stimulation ofJUN expression in GLI1HaCaT and response to signaling caused by tumor promoters, stress, GLI2actHaCaT cells was also observed on the protein UV irradiation and wounding. It is important in level as shown by western blot (Figures 1c and d) and N- tumorigenesis by influencing cell proliferation, survival, terminal phosphorylation at Ser63 ofJUN protein is transformation and invasion (reviewed in Eferl and detectable in both cell lines, suggesting that transcription- Wagner, 2003; Johnson and Nakamura, 2007; Maeda ally active JUN is present (Smeal et al., 1992). In and Karin, 2003; Yates and Rayner, 2002; Zenz and agreement with the qRT–PCR results, JUN protein Wagner, 2006). In mice the effects ofloss ofJun or loss of expression is weaker and delayed in GLI1HaCaT com- Jun activation are highly dependent on cell type and pared to GLI2actHaCaT cells, where a distinct signal is tissue (reviewed in Jochum et al., 2001). Jun is essential in already detectable 12 h after induction. To exclude that embryonic development (Johnson et al., 1993), though JUN induction by GLI is an unusual response restricted to knocking out Jun in keratinocytes results in normal skin the HaCaT cell line, we retrovirally transduced the human architecture. Cultured primary JunÀ/À keratinocytes, keratinocyte cell line N/TERT-1 (Dickson et al., 2000) and however, show reduced proliferation (Zenz et al., 2003). also primary human foreskin keratinocytes (pFSK) with An important contribution ofJUN to cellular prolifera- either enhanced green fluorescent protein (EGFP) or tion is the transcriptional activation ofcyclin D1 human GLI2act. At 96 h after infection, elevated levels of (CCND1) in keratinocytes and other positive regulators JUN mRNA and protein were detectable for GLI2act ofcell-cycle progression as well as inhibition ofnegative expression compared to the EGFP expressing control in regulators (reviewed in Zenz and Wagner, 2006). The both cell types (Figures 1e and f). importance ofJun in tumor initiation was shown in a Having shown that expression ofGLI1 and GLI2act tumor prone mouse model. Papillomas form in mice in cultured keratinocytes leads to an increase in JUN overexpressing the activated son ofsevenless (SOS) gene mRNA and protein, we investigated JUN expression in under the keratin 5 promoter leading to Jun activation BCC, which shows strong expression ofHH pathway through the Ras pathway (Sibilia et al., 2000). In components PTCH, GLI1 and GLI2 (Hahn et al., 1996; contrast, only small tumors formed in the absence of Johnson et al., 1996; Ghali et al., 1999; Regl et al., 2002). Jun (Zenz et al., 2003) and tumor formation was also We measured the mRNA levels ofJUN, GLI1, GLI2 impaired in mice expressing mutant Jun proteins that and PTCH by qRT–PCR in samples ofhuman BCC cannot be phosphorylated and activated by JNK (n ¼ 6) and normal human skin (n ¼ 3) (Figure 2a). BCC (Behrens et al., 1999, 2000). samples showed characteristic high expression ofGLI1, Here we show that JUN is a direct target ofGLI1 and GLI2 and PTCH and also significantly increased JUN GLI2 in human keratinocytes. Cooperative activation of mRNA levels compared to healthy skin (Figure 2a). JUN by GLI and JUN is strongly dependent on one Immunohistochemical staining ofJUN protein on GLI-binding site and the presence ofthe known AP-1 sections ofparaffin-embedded human BCC samples sites. Physical interaction between JUN and GLI2 confirmed qRT–PCR results (Figure 2b). Strong and depends on JNK-mediated phosphorylation ofJUN. specific nuclear staining was found throughout the Decreased JUN expression reduces the cell-cycle-pro- tumor islands pointing to transcriptionally active JUN moting activity ofGLI and a subset ofits target genes protein in accordance with previous studies (Zhang that are jointly regulated by HH/GLI and JUN. et al., 2006). No staining was found in the surrounding stroma. These results support regulation ofJUN expression by HH/GLI signaling in vivo. Results

JUN expression is increased in response to GLI in human The JUN promoter contains at least one functional keratinocytes GLI-binding site We investigated the inﬂuence ofHH/GLI signaling on Because JUN transcription is activated by GLI, we the expression ofAP-1 family members in human searched for putative GLI-binding sites in the promoter

Oncogene Cooperative target gene regulation by GLI and JUN S Laner-Plamberger et al 1641

Figure 1 JUN expression is upregulated by GLI transcription factors. (a and b) qRT–PCR analysis ofJUN ( a) or PTCH (b) mRNA levels in HaCaT keratinocytes stably expressing GLI1 (GLI1HaCaT) or GLI2act (GLI2actHaCaT) under doxycycline (dox) control for the times indicated. Fold change refers to the mRNA ratios of induced to uninduced cells. Data show mean values of three independent experiments using two different cell clones of stably expressing GLI1HaCaT or GLI2actHaCaT cells. (c and d) GLI and JUN protein levels in GLI1HaCaT (c) or GLI2actHaCaT (d) cells were analysed by western blot using antibodies against GLI1 or GLI2 and total JUN or phosphorylated JUN (pJUN). b-Actin (b-actin) was used as a loading control (c and d, bottom lane). (e and f) Western blot (left panels) and qRT–PCR (right panels) analysis of JUN expression in N/TERT-1 (e) or primary human foreskin keratinocytes (pFSK) (f) retrovirally transduced with GLI2act (LL3.7-GLI2act) or enhanced green ﬂuorescent protein (EGFP; LL3.7) as control. Fold change refers to the mRNA ratio of GLI2act to EGFP expressing cells.

region ofhuman JUN. Using ScanACE (Roth et al., potential compared to the N-terminally truncated form 1998) we identified a cluster offive potential GLI- GLI2act (Roessler et al., 2005), activates transcription binding sites (GLIBS1-5) located between À785 and to a similar level as GLI1. The N-terminally truncated À503 upstream ofthe transcription start site (according form of GLI3 (GLI3act), previously described as an to NM_002228) and one isolated potential binding site activator (Sasaki et al., 1999), did not induce detectable at position þ 332 (BS6). All binding sites differ from the luciferase transcription (Figure 3b). To assess the consensus sequence (Kinzler and Vogelstein, 1990) by importance ofthe predicted GLI-binding sites two nucleotide substitutions (Figure 3a). To test these (Figure 3a), all sites were inactivated individually or in putative GLI-binding sites, we cloned a 1243 bp frag- different combinations by mutation at essential posi- ment (À829 bp to þ 414 bp) ofthe JUN promoter tions (see Materials and methods section) (Figure 3c). comprising the predicted GLI-binding sites and the Only mutation ofbinding site 3 (GLIBS3) abolished known AP-1-binding sites JUNBS1 and JUNBS2 reporter gene activity (Figure 3c), whereas mutation (Angel et al., 1988b; Stein et al., 1992) in a luciferase ofthe binding sites 1 (GLIBS1m), 2 (GLIBS2m), 4 reporter vector (JUNprom) (Figure 3a). Transcription (GLIBS4m), 5 (GLIBS5m) or 6 (GLIBS6m) alone or in of luciferase from this reporter can be activated by GLI1 different combinations (GLIBS1to4m, GLIBS3to4m, and GLI2act (Figure 3b). Consistent with qRT–PCR GLIBS1to5m or GLIBS1to6m) (Figure 3c) did not and western blot results (Figure 1), activation by GLI1 result in significant reduction ofpromoter activity. We is significantly weaker compared to GLI2act. Full- therefore conclude that BS3 is essential for transcrip- length GLI2, which is known to have reduced activation tional activation ofthe JUN promoter by GLI.

Oncogene Cooperative target gene regulation by GLI and JUN S Laner-Plamberger et al 1642

Figure 2 JUN is expressed in basal cell carcinoma (BCC). (a) qRT–PCR ofhuman BCCs ( n ¼ 6) and normal human skin samples (n ¼ 3) demonstrating that JUN, GLI1 and GLI2 are expressed in BCC. Data represent mean values ofall tested BCC samples. (b) Immunostaining ofsections ofhuman BCC with a JUN-speciﬁc antibody. A representative sample is shown.

Figure 3 The JUN promoter contains six potential GLI-binding sites and is activated by GLI proteins. (a) Schematic overview ofthe cloned 1243 bp fragment of the human JUN promoter (JUNprom). Activator protein 1 (AP-1)-binding sites JunBS1 and JunBS2 are indicated by circles and predicted GLI-binding sites by boxes. Sequences are listed below. Bases different from the GLI consensus sequence are in bold letters, the orientation ofthe binding site is indicated by ‘ þ ’(50–30)or‘À’(30–50). Positions ofthe GLI-binding sites refer to the transcription start site according to NM_002228. (b) Luciferase reporter assay with the cloned JUN promoter fragment. HaCaT cells were co-transfected with JUNprom and GLI1, GLI2act, GLI2 and GLI3act expression constructs or the empty expression vector p4TO. Data shown are mean values ofrelative light units (RLU) ofthree independent experiments. ( c) Luciferase reporter assay ofJUN promoter constructs with mutated GLI-binding sites (marked by ‘m’) shows that activation by GLI depends on GLIBS3 only. Mutated JUN promoter constructs were co-transfected with GLI2act or empty expression vector (p4TO). The data shown are mean values ofRLU ofthree independent experiments. ( d) Electrophoretic mobility shift assay (EMSA) demonstrating that GLIBS3 is specifically bound by the DNA-binding domain ofGLI2. Specificity ofbinding was shown by competing with increasing amounts ofunlabeled oligonucleotides corresponding to either the wild-type binding sequence GLIBS3, the GLI consensus sequence (Cons), mutated consensus sequence (mut) or the unspecific competitor poly dI-dC.

Oncogene Cooperative target gene regulation by GLI and JUN S Laner-Plamberger et al 1643 Because BS3 differs considerably from the published competitor poly dI-dC (Figure 3d). A multiple sequence GLI consensus binding site, we tested the binding ofa alignment shows that GLIBS3 is completely conserved recombinant purified GLI zinc-finger domain protein to in primates (data not shown), cattle and cat, rat and BS3 in an electrophoretic mobility shift assay (EMSA) mouse share two substitutions (Supplementary Figure (Figure 3d). The GLI zinc-finger domain specifically S5). In addition mouse misses one nucleotide and it is binds to BS3 in this assay as shown by competition with conceivable that this might be a sequencing error excess ofBS3 itselfor the consensus sequence (Cons), (Supplementary Figure S5). All potential GLI-binding but not with mutated oligonucleotide (mut) or unspecific sites shown to be nonfunctional (Figure 3c) are not conserved. Together these results identified BS3 as a functional GLI-binding site that is essential for activation ofJUN by GLI in human keratinocytes.

Jun promoter activity is strongly enhanced by combinatorial action of GLI and JUN itself The fact that JUN is a target of itself and also of GLI implies an interaction between those factors. Using JUN promoter luciferase reporter constructs carrying mutations in either GLIBS3 (Figure 3c) or the JUN-binding sites JUNBS1 and JUNBS2 (Figure 3a), we found that activation ofthe JUN promoter by GLI1 and GLI2act not only depends on the presence ofGLIBS3 but also on both JUN sites. Mutation ofJUNBS1 and JUNBS2 (JUN_JUNBSm, Figure 4a) strongly reduces the activation potential ofboth GLI transcription factors compared to the wild-type JUN promoter (JUNprom) and, as expected, abolished the activation by JUN. In contrast, mutation of BS3 (JUN_GLIBS3) only affects the activation ofthe JUN promoter by GLI but not by JUN (Figure 4a). The cooperative interaction ofJUN and GLI was conﬁrmed using UV irradiation to activate endogenous JUN in HaCaT cells. UV light has been reported to activate JNK leading to hyperphosphorylation and activation ofJUN (Hibi et al., 1993; Figure 4b, inset). We therefore irradiated HaCaT cells co-transfected with GLI1 or GLI2act and JUN promoter

Figure 4 The JUN promoter is cooperatively activated by GLI and JUN. (a) Luciferase reporter assay showing that full activation of the JUN promoter reporter by GLI activators requires both GLI- and JUN-binding sites. HaCaT cells were co-transfected with GLI or JUN expression vectors as indicated, together with either wild-type JUN reporter (JUNprom), a construct carrying mutations in the two functional JUN-binding sites JUNBS1 and JUNBS2 (JUN_JUNBSm) or a reporter construct with the mutated GLI-binding site GLIBS3 (JUN_GLIBS3m, a). Data are mean values ofrelative light units (RLU) ofthree independent experiments. ( b) Hyperphosphorylation of JUN by UV light irradiation strongly enhances GLI activation potential. Luciferase reporter constructs used in (a) were co-transfected with GLI1 or GLI2act expression constructs or empty vector (p4TO) in HaCaT cells. At 48 h after transfection cells were exposed to a short UV light pulse and reporter activity was measured after additional 2 h ofcultivation and compared to UV untreated samples. The phosphorylation state ofJUN was controlled by western blot using a phospho-specific JUN antibody (pJUN) (inset). (c) Chromatin immunoprecipitation (ChIP) demonstrates specific binding ofGLI1 or GLI2 and JUN to the JUN promoter. Chromatin isolated from GLI1HaCaT or GLI2actHaCaT cells was precipitated with either specific antibody (aGLI1, aGLI2, aJUN) or species matched normal IgG (nIgG). Signals for DNA fragments spanning the functional GLI- binding site (GLIBS) or activator protein 1 (AP-1)-binding sites (JUNBS) as indicated in the drawing were only detected in the specific precipitates. No amplification was observed in controls (nIgG) and for a 284 bp fragment of the human RPLP0 promoter.

Oncogene Cooperative target gene regulation by GLI and JUN S Laner-Plamberger et al 1644 Table 1 qRT–PCR analysis ofGLI2/JUN target gene expression hairpin RNA (shRNA) approach (Supplementary Fig- in GL2actHaCaT cells ure S2). mRNA samples ofGLI2actHaCaT cells stably HUGO name +GLI2act ÀGLI2act +GLI2act ÀGLI2act transduced with ctrlshRNA or specific JUNshRNA and +ctrlshRNA +ctrlshRNA +JUNshRNA +JUNshRNA induced for 72 h ( þ GLI2act) or uninduced controls (ÀGLI2act) were analysed on cDNA filter arrays (Regl Group I PTCHa 7.93 1.11 6.86 1.00 et al., 2004a; Eichberger et al., 2006). GLI2act transgene BCL2a 16.03 1.06 16.03 1.00 mRNA and protein levels were unaffected by the GLI1a 3.86 1.04 5.21 1.00 JUNshRNA as determined by qRT–PCR (Table 1, GLI2 21.11 0.65 25.81 1.00 group I) and western blot (Supplementary Figure S2). Selected genes regulated by both GLI2act and JUN Group II MRAS 523.66 3.22 167.44 1.00 were further validated by qRT–PCR (Table 1). We IL1R2a 505.83 19.06 17.69 1.00 found that JUN expression is required for the upregula- ARCa 396.86 2.22 52.26 1.00 tion ofa subset ofGLI target genes only (Table 1, group MEOX1 170.37 21.22 8.77 1.00 II) whereas expression ofcanonical HH/GLI target WNT11 82.71 6.21 35.26 1.00 CSF2b 79.07 3.33 10.41 1.00 genes PTCH, GLI1 and BCL2 is not influenced by TGM3b 66.14 0.92 23.30 1.00 JUNshRNA in induced samples ( þ GLI2act) (Table 1, SFRP1 53.72 2.19 15.03 1.00 group I). There is striking cooperation between GLI2act JUNb 25.68 2.71 2.63 1.00 and JUN for a number of genes, among them MRAS, TCEA2 11.61 1.16 2.88 1.00 IL1R2, ARC, MEOX1, CSF2, TGM3, TCEA2, S100A7a 9.29 7.30 2.27 1.00 E2F1 9.00 2.43 3.40 1.00 HTRA1 and JUN itself(Table 1, group II). GLI2act- HTRA1 8.20 1.91 1.96 1.00 dependent mRNA expression ofthese genes is strongly IRF1 7.27 2.70 1.94 1.00 reduced in the absence ofJUN ( þ GLI2act/ BIK 6.29 2.96 2.19 1.00 þ JUNshRNA compared to þ GLI2act/ þ ctrlshRNA). CCND1a,b 5.90 2.20 1.52 1.00 MMP2 5.81 0.85 2.20 1.00 For other genes activation in the presence ofJUN and FYN 5.08 1.57 2.54 1.00 GLI may be additive as shown for WNT11, CCND1, PSMB9 3.88 2.11 2.06 1.00 IKBKE and FYN (Table 1, group II). To extend these DVL1 3.67 0.93 1.68 1.00 findings to non-HaCaT keratinocytes, we used N/ IKBKE 3.29 2.89 1.13 1.00 TERT-1 keratinocytes that are reported to have growth ETS2 3.37 1.52 1.78 1.00 and differentiation characteristics similar to primary Values represent relative mRNA levels over RPLP0 control gene keratinocytes (Dickson et al., 2000). Cells were trans- expression normalized to the ÀGLI2act/+JUNshRNA sample. duced with LL3.7 lentivirus expressing EGFP-tagged +GLI2act ¼ +DOX; ÀGLI2act ¼ÀDOX. GLI2act (LL3.7-GLI2act) or EGFP (LL3.7) (control) aDescribed as direct GLI target gene (Agren et al., 2004; Ikram et al., (Kasper et al., 2007) followed by retroviral transduction 2004; Kasper et al., 2006a; Regl et al., 2004b). bDescribed as AP1/JUN target gene (Angel et al., 1988b; Gramigni with JUNshRNA or ctrlshRNA. mRNA levels of et al., 1998; Hayakawa et al., 2004; Herber et al., 1994). selected genes from Table 1 were analysed by qRT- PCR (Figure 5). As in HaCaT cells, knockdown ofJUN results in strongly reduced expression ofGLI2act/JUN constructs to efficiently induce JUN activation target genes without significant influence on canonical (Figure 4b). UV irradiation in the absence ofGLI leads HH target genes such as GLI1 (Figure 5). Whether all to a twofold increase in promoter activity (JUNprom) genes listed in Table 1 are directly regulated by GLI regardless ofthe presence ofGLIBS3 (JUN_GLIBS3m) cannot be deduced from the qRT–PCR data, though for whereas UV irradiation has no effect in the absence of some, direct regulation by GLI and JUN was shown by JUNBS (JUN_JUNBSm, Figure 4b). When the cells are luciferase reporter assays (Supplementary Figure S3). irradiated in the presence ofGLI, a dramatic increase in Others have previously been described as direct GLI wild-type promoter (JUNprom) activity was observed targets (footnote a, Table 1) or direct AP-1 targets demonstrating a synergistic interaction between GLI (footnote b, Table 1). A similar experiment was and JUN. Chromatin immunoprecipitation (ChIP) conducted in HaCaT cells expressing GLI1 (GLI1Ha- extended this result to the physical presence ofthe two CaT) and some genes selected from Table 1, group II transcription factors on the JUN promoter (Figure 4c). also showed cooperative activation though quantitative In GLI1HaCaT (upper panels) and GLI2actHaCaT differences to GLI2act were observed (Supplementary (lower panels) cells, JUN and GLI were found at their Table S5). In summary, these data point to cooperative respective binding sites as demonstrated by specific regulation between GLI and JUN on a subset of precipitation ofDNA fragments spanning either HH/GLI target genes in epidermal cells, suggesting a GLIBS3 (fragment GLIBS) or the two JUN-binding novel mechanism ofmodulating HH/GLI target gene sites (fragment JUNBS). expression.

Modulation of GLI target gene expression by JUN Physical interaction between GLI2 and JUN requires We next asked whether other GLI target genes are phosphorylation of JUN regulated similarly. We therefore stably knocked down Cooperative target gene regulation may be associated JUN in GLI2actHaCaT cells by a retroviral short- with physical interaction between GLI and JUN.

Oncogene Cooperative target gene regulation by GLI and JUN S Laner-Plamberger et al 1645 differences in the interaction of GLI1 and GLI2act with JUN.

The positive effect of GLI on cell-cycle progression is dependent on JUN Recently, GLI1 and GLI2 were shown to promote cell- cycle progression and antagonize contact inhibition in HaCaT keratinocytes (Regl et al., 2002). We found activation ofcell-cycle markers such as cyclin D1 and E2F1 by GLI2act and JUN (Table 1), which prompted us to analyse whether GLI can stimulate S-phase entry in epidermal cells in the absence ofJUN. We induced transgene expression in GLI2actHaCaT cells that were retrovirally transduced with either ctrlshRNA or Figure 5 Cooperative GLI2/JUN target gene regulation in N/ TERT-1 cells. qRT–PCR analysis ofIL1R2, ARC, MEOX1, CSF2 JUNshRNA and determined bromodeoxyuridine and GLI1 (control) mRNA levels in N/TERT-1 keratinocytes (BrdU) incorporation at 48 h after confluence as an retrovirally transduced with GLI2act or enhanced green fluorescent indicator ofcell proliferation(Figure 7). Protein levels protein (EGFP) (control) followed by infection with JUNshRNA ofGLI2 transgene and endogenous JUN were analysed and ctrlshRNA virus. Fold change refers to the mRNA ratio of by western blot (Figure 7c). In the presence ofJUN GLI2act to EGFP-expressing cells transduced with either JUNshR- NA (grey bars) or ctrlshRNA (black bars). protein ( þ ctrlshRNA), induction ofGLI2act expression led to a strong increase in BrdU incorporation from a basal level ofapproximately 10% BrdU-positive cells (ÀGLI2act/ þ ctrlshRNA) up to 30% ( þ GLI2act/ Co-immunoprecipitation (CoIP) ofGLI1 or GLI2act þ ctrlshRNA) (Figures 7a and b). In contrast, the cell- and JUN protein from inducible HaCaT cells expressing cycle-promoting activity ofGLI2act is significantly MYC-tagged GLI1 (MGLI1HaCaT) or MYC-tagged reduced upon JUN knockdown ( þ GLI2act/ GLI2act (MGLI2actHaCaT) detected interaction of þ JUNshRNA) (Figures 7a and b), suggesting that endogenous JUN with GLI2act but not GLI1 cooperative activation ofcell-cycle genes by GLI and (Figure 6a). We confirmed this surprising observation JUN leads to enhanced cell proliferation that might in transiently transfected human embryonic kidney have consequences for tumor formation and growth. (HEK) cells. HEK cells were co-transfected with either MYC-tagged GLI1 (MGLI1) or MYC-tagged GLI2act (MGLI2act) together with FLAG-tagged JUN (FJUN) expression constructs or empty expression vector Discussion (pCMV). FLAG-tagged SUFU (FSUFU), which should interact with both GLI1 and GLI2, was used as positive Expression ofGLI transcription factorsin mouse control (Pearse et al., 1999). Again, we detected only epidermis can result in BCC whereas activation ofJun interaction ofJUN and GLI2act but not ofJUN and is required for papilloma formation in mouse skin GLI1 (Figure 6b), although both GLI1 and GLI2act (Young et al., 1999; Behrens et al., 2000). Though the showed specific interaction with SUFU. Because phospho- effects of these two oncogenes on skin are diverse, we rylation ofJUN by JNK is important in protein–protein show here a novel interaction between JUN and the GLI interactions (Bannister et al., 1995), we investigated the transcription factors leading to cooperative target gene effect of JUN phosphorylation on the interaction of regulation in human keratinocytes. JUN itselfis a direct JUN with GLI2act by CoIP in the presence ofSP600125 transcriptional target ofGLI1 and GLI2, though a potent inhibitor ofJNK. Interaction ofJUN and GLI2act is a much more efficient activator of JUN GLI2act was strongly diminished, although protein transcription than is GLI1. This is not due to a generally levels in the input were unchanged compared to controls higher activation potential as PTCH activation is (Figures 6c and d; Supplementary Figure S4). This comparable. Differences in target gene activation was observed in transiently transfected HEK cells between GLI1 and GLI2act have previously been shown (Figure 6d) as well as in MGLI2actHaCaT cells for other promoters (Regl et al., 2004b; Zhao et al., (Figure 6c). The ability ofphosphorylated JUN to 2006; Eichberger et al., 2008). GLI2act that shows interact with cofactors in MGLI2actHaCaT cells was enhanced transcriptional activation compared to full- verified using CBP, which has previously been shown to length GLI2 (Roessler et al., 2005) has been used as an bind JUN in a phosphorylation-dependent manner ‘activated GLI2’ in vivo and in vitro. There is no (Bannister et al., 1995). CBP coprecipitated with JUN evidence for the existence of the protein, though recently in the absence ofSP600125 but not in its presence Speek et al. (2006) identified a GLI2 splice variant (Figure 6c). To exclude a general effect of SP600125 on missing the N-terminal repressor domain in human GLI protein interactions, we used SUFU as control in gonadal tissue and several cell lines with comparable HEK cells (Figure 6d). These results indicate that the transcriptional activity in vitro. interaction ofGLI2act and JUN depends on the The JUN promoter contains at least one functional phosphorylation ofJUN by JNK, whereas suggesting GLI-binding site as shown by mutation and consequent

Oncogene Cooperative target gene regulation by GLI and JUN S Laner-Plamberger et al 1646

Figure 6 GLI2act physically interacts with JUN. (a) Co-immunoprecipitation (CoIP) ofGLI1 or GLI2act and JUN with either anti- MYC (aMYC) or an anti-JUN antibody (aJUN) from cells treated with doxycycline for 72 h. A specific interaction was only detected for GLI2act and JUN (left lower panel, lanes 2 and 3) but not for GLI1 and JUN (right lower panel, lanes 2 and 3). Normal mouse IgG (mIgG) that was used in all immunoprecipitations as control for unspecific interactions gave no signals (right and left lower panels, lane 1). Western blot ofinput proteins ofthe transgenes MGLI1 or MGLI2act and endogenous JUN in MGLI2actHaCaT or MGLI1HaCaT cells are shown in the upper panels. (b) CoIP ofGLI proteins and JUN in HEK293FT cells transiently transfectedwith the indicated expression constructs. Specific interaction between GLI2act and JUN is also present in human embryonic kidney (HEK) cells (left lower panels, lanes 1 and 4) whereas again, no interaction was detected for GLI1 and JUN (right lower panels, lanes 1 and 4). Proteins were precipitated with anti-FLAG agarose (FLAG) and detected by western blot with anti-MYC or anti-FLAG antibodies or a specific JUN antibody as indicated. Upper panels show protein levels in the input, and lower panels show precipitated proteins. FLAG-tagged SUFU was used as a positive control (left and right lower panels, lane 2) and the empty expression vector pCMV-FLAG (pCMV) as negative control (left and right lower panels, lane 3). (c and d) Interaction ofGLI2act with JUN requires N-terminal phosphorylation ofJUN. ( c) MGLI2actHaCaT cells were induced for 24 h with doxycycline and treated with JNK inhibitor SP600125 or dimethyl sulfoxide (DMSO). CoIP of MGLI2act with endogenous JUN protein was carried out with a specific JUN antibody (aJUN). As control, normal mouse IgG (mIgG) was used. Western blot showed that interaction between JUN and MGLI2act can only be detected in DMSO-treated samples (lane 1). As positive control for a phosphorylation-dependent JUN interaction, we used CBP (lane 1). (d) Comparable results were obtained in transiently transfected HEK293FT cells (lanes 2 and 5). Cells were transfected with expression constructs as indicated. For precipitation, anti-FLAG agarose (FLAG) was used. To exclude a general effect of SP600125 on GLI2 protein interaction, we used FLAG-tagged SUFU (FSUFU) as positive control (lanes 1 and 4). Input protein levels are shown in upper panels, and lower panels show immunoprecipitated proteins.

Oncogene Cooperative target gene regulation by GLI and JUN S Laner-Plamberger et al 1647

Figure 7 Reduced BrdU incorporation in GLI expressing keratinocytes treated with JUNshRNA. (a and b) BrdU incorporation showing that GLI2act-induced G1/S-phase progression requires JUN activity. GLI2actHaCaT were induced with doxycycline ( þ GLI2act) and transduced with JUNshRNA or ctrlshRNA. (b) Ratio ofBrdU-positive cells to total cell number (DAPI staining). Bars show the mean values oftwo independent experiments performedin triplicate. ( c) Western blot showing GLI2act (GLI2act) and JUN protein (JUN) levels. b-Actin is shown as loading control (b-actin). loss ofactivation potential by GLI. When the AP-1 sites recruited on the JUN promoter in the expected region are mutated the activation by GLI is strongly reduced, and so is JUN. Physical interaction between JUN and whereas mutation of the GLI site does not affect relative GLI2 was detected by CoIP though surprisingly, no activation by JUN. The requirement for JUN in GLI- interaction was seen with GLI1. This may indicate that dependent promoter activation is also shown by the interaction is too weak or transient to be observed shRNA-mediated JUN knockdown and through UV by this method or that the lower activation potential of activation ofendogenous JUN. In the latter case, the GLI1 on the JUN promoter is due to differences in the absolute level oftranscription by JUN itselfis greatly activation complex. increased by concomitant GLI expression, revealing JUN has been shown to interact with many other strong synergism between the two transcription factors. transcription factors and is a member of the well- This is similar to interaction ofTCF4 and JUN, also on characterized enhanceosome complex at the IFN-b the JUN promoter, as reported by Nateri et al. (2005). promoter (Maniatis et al., 1998) where ATF-2/JUN, The presence ofthe TCF- and JUN-binding sites greatly IRF-3 and IRF-7, and NF-kB (p50/RelA) are necessary enhances activation by JUN and TCF4, whereas in the for promoter activation. Several members of the absence ofthe JUN sites activation by TCF4 is complex including JUN are able to interact with CBP abolished (Nateri et al., 2005). TCF4 and JUN have or p300. The binding sequences for these transcription both been shown to bind directly the JUN promoter and factors are linearly arranged, in proximity or over- interact physically. Similarly, GLI1 and GLI2 are lapping on the DNA whereas relatively few protein–

Oncogene Cooperative target gene regulation by GLI and JUN S Laner-Plamberger et al 1648 protein interactions are observed in the crystal structure fetal calf serum (PAA), 100 mg/ml streptomycin and 62.5 mg/ml (Panne, 2008). JUN also frequently interacts with penicillin (Invitrogen, Carlsbad, CA, USA) at 37 1C, 5% CO2. members of the ETS transcription factor family, which N/TERT-1 cells were grown in Keratinocyte Serum-free usually requires very close spacing for interaction to Medium (Invitrogen) with 100 mg/ml streptomycin and occur (Basuyaux et al., 1997; Kim et al., 2006). A small 62.5 mg/ml penicillin, pFSK were cultivated in EpiLife basal keratinocyte medium (Invitrogen) supplemented with growth distance between sites has also been reported for JUN– supplement (HKGS) (Invitrogen), 5% fetal calf serum (PAA), STAT3 interaction (Schaefer et al., 1995; Ginsberg 100 mg/ml streptomycin and 62.5 mg/ml penicillin. et al., 2007). It appears that such close spacing is not The T-REx System (Invitrogen) was used to generate required for the functional interaction of TCF4 with MGLI1HaCaT and MGLI2actHaCaT double-stable cell lines JUN on the JUN promoter (Nateri et al., 2005). In the expressing NLS-MYC-tagged human GLI1 or NLS-MYC- case ofthe MYC and cyclin D1 promoters, an tagged human GLI2act. For double-stable inducible HaCaT interaction ofJUN, b-catenin and TCF4 does not even lines MGLI1HaCaT or MGLI2actHaCaT and GLI1HaCaT require an AP-1 site (Toualbi et al., 2007). Similar to the (Eichberger et al., 2006) or GLI2actHaCaT (Regl et al., 2004a) interaction ofTCF4 with JUN, interaction ofGLI with medium was supplemented with 25 mg/ml zeocin (Invitrogen) JUN does not require close spacing, not only in the JUN and 8 mg/ml blasticidin S (Sigma-Aldrich, St Louis, MO, USA). Transgene expression was induced by 50 ng/ml dox- promoter itself, but also in the other characterized JUN/ ycycline (Sigma-Aldrich). GLI-dependent promoters (Supplementary Figure S3; Supplementary Table S6). Retroviral gene expression and shRNA-mediated knockdown GLI2act itselfdoes not necessarily interact directly Lentiviral vectors used for expression of EGFP (LL3.7) with JUN but may be a member ofa larger complex that (Rubinson et al., 2003) and EGFP-tagged GLI2act (LL3.7- includes CBP, a known partner ofJUN (Bannister et al., GLI2act) are described in Kasper et al. (2007). 1995) and possibly other interactors ofJUN. Physical For shRNA-mediated JUN knockdown, a single shRNA interaction between JUN and GLI2, however, depends (TRCN0000039590) from the lentiviral MISSION shRNA set on N-terminal phosphorylation ofJUN because it SHGLY-N2M_002228 (Sigma-Aldrich) and a nontarget con- cannot be observed when the samples are treated with trol shRNA (SHC002) were selected. Virus production and the JNK inhibitor SP600125. This implies the formation infection of cells was performed essentially as described ofa transcriptional activation complex enabled by in Kasper et al. (2007). At 24 h after virus infection, medium phosphorylation ofJUN, which then leads to coopera- was supplemented with 1 mg/ml puromycin (Sigma-Aldrich) to select for infected cells. Doxycycline (Sigma-Aldrich) tive activation ofGLI/JUN target genes. Because JUN treatment (50 ng/ml medium) for transgene expression in is activated in response to many stimuli such as the GLI2actHaCaT cells was started 72 h after infection. EGFR signal that has been shown to induce IL1R2 synergistically with GLI (Kasper et al., 2006b; Aberger Cell proliferation assay F, personal communication), it is tempting to speculate BrdU incorporation assays in shRNA-transduced GLI1Ha- that JUN may act as an integrator ofmultiple inputs CaT or GLI2actHaCaT were carried out as described (Regl and is modulating the expression ofHH/GLI target et al., 2002), using the FLUOS in situ cell proliferation kit genes. Furthermore, our results show that JUN is also (Roche, Basel, Switzerland). BrdU incorporation was detected required for the cell-cycle-promoting activity of GLI in with an Alexa Fluor 594-conjugated anti-BrdU mouse anti- keratinocytes, suggesting that the JUN-GLI interaction body (Invitrogen). Microscopic imaging was carried out on an may have important consequences for HH-dependent Olympus IX 70 microscope equipped with a SPOT CCD skin tumor formation. camera (Diagnostic Instruments Inc., Sterling Heights, MI, USA). BrdU-positive cells and 4,6-diamidino-2-phenyl indole (DAPI)-stained nuclei were counted from two independent experiments carried out in triplicate. Materials and methods RNA isolation, high-density DNA filter array production Cloning of promoter and expression constructs and qRT–PCR analysis For the JUN promoter reporter plasmid (JUNprom), a Total RNA from BCC and normal skin was isolated with TRI 1243 bp fragment of human JUN (NM_002228) was amplified Reagent (Molecular Research Center Inc., Cincinnati, OH, by PCR (for primers see Supplementary Table S1) from USA) followed by LiCl precipitation. Total RNA of HaCaT, human genomic DNA, digested with BlpI/XhoI and cloned N/TERT-1 and pFSK cells was isolated and purified with the into the luciferase reporter vector pGL3basic (Promega, High Pure RNA Isolation Kit (Roche). cDNA synthesis and Madison, WI, USA). Potential GLI-binding sites were qRT-PCR analysis were carried out as described in Eichberger mutated at positions 6 and 7 to G and AP-1-binding sites et al. (2006). Human large ribosomal protein P0 (RPLP0) was JUNBS1 and JUNBS2 were mutated as described in Angel used for normalization of sample material in qRT–PCR et al et al . (1988b) and Stein . (1992) using the QuickChange analysis (Martin et al., 2001). For primer sequences see site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA). Supplementary Table S2. For expression profiling on cDNA GLI1, GLI2, GLI2act, GLI3act, JUN, SUFU expression filter arrays containing 2135 EST clones (Regl et al., 2004a), et al constructs have been described previously (Schutte ., 1989; cells were harvested after 72 h of doxycycline treatment, total et al et al Kogerman ., 1999; Eichberger ., 2008). mRNA was isolated and purified according to manufacturer’s protocols (High Pure RNA Isolation Kit; Roche) and cDNA Cell culture synthesized from 15 mg total RNA with Superscript II (RNase HaCaT cells were cultured in Dulbecco’s modified Eagle’s HÀ) reverse transcriptase (Invitrogen) using oligo dT primers, medium (high glucose; PAA, Pasching, Austria) with 10% according to manufacturer’s instructions. 33P-labeling ofthe

Oncogene Cooperative target gene regulation by GLI and JUN S Laner-Plamberger et al 1649 cDNA, array production, hybridization and analysis were rabbit-anti-CBP (CBP-A22), all from Santa Cruz Biotechnol- carried out as described earlier (Eichberger et al., 2006). ogy. Control reactions were performed with species matched normal IgGs (Santa Cruz Biotechnology). PCR primer Western blot and immunohistochemistry sequences are listed in Supplementary Table S4. Cells were lysed in 125 mM Tris (pH 6.8), 5% glycerol, 2% SDS, 1% b-mercaptoethanol, 0.006% bromophenol blue and Co-immunoprecipitation proteins resolved by SDS–polyacrylamide gel electrophoresis For the detection ofGLI1 or GLI2act protein interactions, (SDS–PAGE). GLI and JUN were detected using the MGLI1HaCaT and MGLI2actHaCaT cells were grown in following primary antibodies: polyclonal goat-anti-GLI1 6-well plates and induced for 72 h. HEK293FT cells were (GLI1-C18), polyclonal rabbit-anti-GLI2 (GLI2-H300), poly- grown in 6-well plates and transfected with expression clonal rabbit-anti-JUN (JUN-H79), monoclonal mouse-anti- construct using Transfectin (Bio-Rad, Hercules, CA, USA) pJUN (p-c-JUN (KM1)) and polyclonal rabbit-anti-CBP according to manufacturer’s instructions. For inhibition of (CBP-A22), all from Santa Cruz Biotechnology (Santa Cruz, JNK, the inhibitor SP600125 (Sigma-Aldrich) was added to a CA, USA). Secondary antibodies were HRP-conjugated goat- final concentration of10 mM for 24 h, controls were treated anti-rabbit, chicken-anti-goat (Santa Cruz Biotechnology) and with DMSO (Sigma-Aldrich). sheep-anti-mouse (GE Healthcare, Little Chalfont, UK). CoIPs were carried out as described in Giesecke et al. (2006) Proteins were visualized using the SuperSignal West detection with some modifications. Briefly all buffers were supplemented system (PIERCE, Rockford, IL, USA). with protease inhibitor cocktail (Sigma-Aldrich) and Phos- Paraffin-embedded sections of human BCC were stained for STOP phosphatase inhibitor cocktail (Roche). Cells from three JUN as described (Eichberger et al., 2008) using a mouse wells were pooled in cold buffer A (10 mM HEPES (pH 7.8), monoclonal antibody directed against JUN (BD Biosciences, 1.5 mM MgCl2,10mM KCl, 20 mM ZnCl2), homogenized and San Jose, CA, USA). lysed in 10 volumes cold buffer B (150 mM NaCl, 50 mM Tris (pH 7.5), 0.1% SDS, 1% NP-40). Lysates were precleared with Luciferase reporter gene assay IgG Sepharose 6 Fast Flow (GE Healthcare) for 1 h HaCaT cells were grown in 24-well plates to 80% confluency, Immunoprecipitation was carried out with 40 ml FLAG- and transfected in triplicate with the respective expression agarose (Sigma-Aldrich) or 4 mg ofthe followingantibodies constructs, shRNA constructs (ctrlshRNA or JUNshRNA) followed by adsorption to protein G Sepharose 4 Fast Flow and pGL3 basic luciferase reporter plasmids. A lacZ expres- (GE Healthcare): monoclonal mouse-anti-MYC (Santa Cruz sion plasmid was co-transfected for normalization. Transfec- Biotechnology), monoclonal mouse-anti-JUN (BD Bio- tion was carried out using SuperFect transfection reagent sciences) and normal mouse IgG as control (Santa Cruz (QIAGEN Inc., Valencia, CA, USA) according to manufac- Biotechnology). Precipitates were washed 3 Â with cold buffer turer’s protocol. Luciferase activity in cell lysates was B and proteins were eluted with 250 mM Tris (pH 6.8), 10% measured 48 h after transfection with a LucyII luminometer glycerol, 4% SDS, 2% b-mercaptoethanol and 0.006% (Anthos Labtec, Cambridge, UK) using Luciferase Assay bromophenol blue and subjected to SDS–PAGE followed by Substrate (Promega) according to manufacturer’s instructions. western blot analysis. UV irradiation ofHaCaT cells was carried out with 750 kJ/m 2 using a Stratalinker 1800 UV crosslinker (Stratagene) 48 h after transfection, after additional 2 h of growth cells were Acknowledgements subjected to the analysis. We thank Dr Fritz Aberger for stimulating discussions and Electrophoretic mobility shift assay and chromatin critical reading of the paper, Mag. Stefan Klingler for advice immunoprecipitation on propagation ofhuman primary foreskinkeratinocytes and EMSA was carried out as described previously (Eichberger Dr Martina Winklmayr for help with co-immunoprecipitations et al., 2008). Oligonucleotides used for EMSA are shown in and cloning. We are particularly grateful to Dr Michael Supplementary Table S3. ChIP from GLI1HaCaT or J Birrer for providing the JUN expression plasmid and Dr GLI2actHaCaT was carried out as described in Eichberger Rune Toftgard for a SUFU expression construct. This work et al. (2008). Antibodies used were monoclonal mouse-anti- was supported by the Austrian Genome Project GENAU JUN (BD Biosciences), polyclonal goat-anti-GLI2 (GLI2- Ultrasensitive Proteomics and Genomics II and the University N20), polyclonal goat-anti-GLI1 (GLI1-C18) and polyclonal ofSalzburg priority program Biosciences and Health.

References

Agren M, Kogerman P, Kleman MI, Wessling M, Toftgard R. (2004). required for CBP induced stimulation in vivo and CBP binding in Expression ofthe PTCH1 tumor suppressor gene is regulated by vitro. Oncogene 11: 2509–2514. alternative promoters and a single functional Gli-binding site. Gene Basuyaux JP, Ferreira E, Stehelin D, Buttice G. (1997). The 330: 101–114. Ets transcription factors interact with each other and with the Angel P, Allegretto EA, Okino ST, Hattori K, Boyle WJ, Hunter T c-Fos/c-Jun complex via distinct protein domains in a et al. (1988a). Oncogene jun encodes a sequence-speciﬁc trans- DNA-dependent and -independent manner. J Biol Chem 272: activator similar to AP-1. Nature 332: 166–171. 26188–26195. Angel P, Hattori K, Smeal T, Karin M. (1988b). The jun proto- Beachy PA, Karhadkar SS, Berman DM. (2004). Tissue repair and oncogene is positively autoregulated by its product, Jun/AP-1. Cell stem cell renewal in carcinogenesis. Nature 432: 324–331. 55: 875–885. Behrens A, Jochum W, Sibilia M, Wagner EF. (2000). Oncogenic Bannister AJ, Oehler T, Wilhelm D, Angel P, Kouzarides T. (1995). transformation by ras and fos is mediated by c-Jun N-terminal Stimulation ofc-Jun activity by CBP: c-Jun residues Ser63/73 are phosphorylation. Oncogene 19: 2657–2663.

Oncogene Cooperative target gene regulation by GLI and JUN S Laner-Plamberger et al 1650 Behrens A, Sibilia M, Wagner EF. (1999). Amino-terminal phosphor- Ingham PW, McMahon AP. (2001). Hedgehog signaling in animal ylation ofc-Jun regulates stress-induced apoptosis and cellular development: paradigms and principles. Genes Dev 15: 3059–3087. proliferation. Nat Genet 21: 326–329. Jochum W, Passegue E, Wagner EF. (2001). AP-1 in mouse Bigelow RL, Jen EY, Delehedde M, Chari NS, McDonnell TJ. (2005). development and tumorigenesis. Oncogene 20: 2401–2412. Sonic hedgehog induces epidermal growth factor dependent matrix Johnson GL, Nakamura K. (2007). The c-jun kinase/stress-activated infiltration in HaCaT keratinocytes. J Invest Dermatol 124: 457–465. pathway: regulation, function and role in human disease. Biochim Dickson MA, Hahn WC, Ino Y, Ronfard V, Wu JY, Weinberg RA Biophys Acta 1773: 1341–1348. et al. (2000). Human keratinocytes that express hTERT and also Johnson RL, Rothman AL, Xie J, Goodrich LV, Bare JW, Bonifas JM bypass a p16(INK4a)-enforced mechanism that limits life span et al. (1996). Human homolog ofpatched, a candidate gene forthe become immortal yet retain normal growth and differentiation basal cell nevus syndrome. Science 272: 1668–1671. characteristics. Mol Cell Biol 20: 1436–1447. Johnson RS, van Lingen B, Papaioannou VE, Spiegelman BM. (1993). Eferl R, Wagner EF. (2003). AP-1: a double-edged sword in A null mutation at the c-jun locus causes embryonic lethality and tumorigenesis. Nat Rev Cancer 3: 859–868. retarded cell growth in culture. Genes Dev 7: 1309–1317. Eichberger T, Kaser A, Pixner C, Schmid C, Klingler S, Winklmayr M Kasper M, Regl G, Eichberger T, FrischaufAM, Aberger F. (2007). et al. (2008). GLI2-specific transcriptional activation ofthe bone Efficient manipulation of Hedgehog/GLI signaling using retroviral morphogenetic protein/activin antagonist follistatin in human expression systems. Methods Mol Biol 397: 67–78. epidermal cells. J Biol Chem 283: 12426–12437. Kasper M, Regl G, FrischaufAM, Aberger F. (2006a). GLI Eichberger T, Sander V, Schnidar H, Regl G, Kasper M, Schmid C transcription factors: mediators of oncogenic Hedgehog signalling. et al. (2006). Overlapping and distinct transcriptional regulator Eur J Cancer 42: 437–445. properties ofthe GLI1 and GLI2 oncogenes. Genomics 87: 616–632. Kasper M, Schnidar H, Neill GW, Hanneder M, Klingler S, Blaas L Evangelista M, Tian H, de Sauvage FJ. (2006). The hedgehog signaling et al. (2006b). Selective modulation ofHedgehog/GLI target gene pathway in cancer. Clin Cancer Res 12: 5924–5928. expression by epidermal growth factor signaling in human Ghali L, Wong ST, Green J, Tidman N, Quinn AG. (1999). Gli1 keratinocytes. Mol Cell Biol 26: 6283–6298. protein is expressed in basal cell carcinomas, outer root sheath Kim S, Denny CT, Wisdom R. (2006). Cooperative DNA binding with keratinocytes and a subpopulation ofmesenchymal cells in normal AP-1 proteins is required for transformation by EWS–Ets fusion human skin. J Invest Dermatol 113: 595–599. proteins. Mol Cell Biol 26: 2467–2478. Giesecke AV, Fang R, Joung JK. (2006). Synthetic protein-protein Kinzler KW, Vogelstein B. (1990). The GLI gene encodes a nuclear interaction domains created by shuffling Cys2His2 zinc-fingers. Mol protein which binds specific sequences in the human genome. Syst Biol 2: 2006.2011 Epub 21 March 2006. Mol Cell Biol 10: 634–642. Ginsberg M, Czeko E, Muller P, Ren Z, Chen X, Darnell Jr JE. (2007). Kogerman P, Grimm T, Kogerman L, Krause D, Unden AB, Amino acid residues required for physical and cooperative Sandstedt B et al. (1999). Mammalian suppressor-of-fused transcriptional interaction ofSTAT3 and AP-1 proteins c-Jun and modulates nuclear-cytoplasmic shuttling ofGli-1. Nat Cell Biol 1: c-Fos. Mol Cell Biol 27: 6300–6308. 312–319. Grachtchouk M, Mo R, Yu S, Zhang X, Sasaki H, Hui CC et al. Lauth M, Bergstrom A, Toftgard R. (2007). Phorbol esters inhibit the (2000). Basal cell carcinomas in mice overexpressing Gli2 in skin. Hedgehog signalling pathway downstream ofsuppressor offused, Nat Genet 24: 216–217. but upstream ofGli. Oncogene 26: 5163–5168. Gramigni C, Penco S, Bianchi-Scarra G, Ravazzolo R, Garre C. Maeda S, Karin M. (2003). Oncogene at last—c-Jun promotes liver (1998). An upstream negative regulatory element in human cancer in mice. Cancer Cell 3: 102–104. granulocyte-macrophage colony-stimulating factor promoter is Maniatis T, Falvo JV, Kim TH, Kim TK, Lin CH, Parekh BS et al. recognised by AP1 family members. FEBS Lett 440: 119–124. (1998). Structure and function of the interferon-beta enhanceosome. Hahn H, Wicking C, Zaphiropoulous PG, Gailani MR, Shanley S, Cold Spring Harb Symp Quant Biol 63: 609–620. Chidambaram A et al. (1996). Mutations ofthe human homolog of Martin KJ, Graner E, Li Y, Price LM, Kritzman BM, Fournier MV Drosophila patched in the nevoid basal cell carcinoma syndrome. et al. (2001). High-sensitivity array analysis ofgene expression for Cell 85: 841–851. the early detection ofdisseminated breast tumor cells in peripheral Hallikas O, Palin K, Sinjushina N, Rautiainen R, Partanen J, blood. Proc Natl Acad Sci USA 98: 2646–2651. Ukkonen E et al. (2006). Genome-wide prediction ofmammalian Mechta-Grigoriou F, Gerald D, Yaniv M. (2001). The mammalian Jun enhancers based on analysis of transcription-factor binding affinity. proteins: redundancy and specificity. Oncogene 20: 2378–2389. Cell 124: 47–59. Mimeault M, Johansson SL, Vankatraman G, Moore E, Henichart JP, Hayakawa J, Mittal S, Wang Y, Korkmaz KS, Adamson E, English C Depreux P et al. (2007). Combined targeting ofepidermal growth et al. (2004). Identification ofpromoters bound by c-Jun/ATF2 factor receptor and hedgehog signaling by gefitinib and cyclopamine during rapid large-scale gene activation following genotoxic stress. cooperatively improves the cytotoxic effects of docetaxel on Mol Cell 16: 521–535. metastatic prostate cancer cells. Mol Cancer Ther 6: 967–978. Hebrok M. (2003). Hedgehog signaling in pancreas development. Nateri AS, Spencer-Dene B, Behrens A. (2005). Interaction of Mech Dev 120: 45–57. phosphorylated c-Jun with TCF4 regulates intestinal cancer Herber B, Truss M, Beato M, Muller R. (1994). Inducible regulatory development. Nature 437: 281–285. elements in the human cyclin D1 promoter. Oncogene 9: 1295–1304. Neill GW, Harrison WJ, Ikram MS, Williams TD, Bianchi LS, Hess J, Angel P, Schorpp-Kistner M. (2004). AP-1 subunits: quarrel Nadendla SK et al. (2008). GLI1 repression ofERK activity and harmony among siblings. J Cell Sci 117: 5965–5973. correlates with colony formation and impaired migration in human Hibi M, Lin A, Smeal T, Minden A, Karin M. (1993). Identification of epidermal keratinocytes. Carcinogenesis 29: 738–746. an oncoprotein- and UV-responsive protein kinase that binds and Nilsson M, Unden AB, Krause D, Malmqwist U, Raza K, potentiates the c-Jun activation domain. Genes Dev 7: 2135–2148. Zaphiropoulos PG et al. (2000). Induction ofbasal cell carcinomas Hooper JE, Scott MP. (2005). Communicating with Hedgehogs. Nat and trichoepitheliomas in mice overexpressing GLI-1. Proc Natl Rev Mol Cell Biol 6: 306–317. Acad Sci USA 97: 3438–3443. Huangfu D, Anderson KV. (2006). Signaling from Smo to Ci/Gli: Panne D. (2008). The enhanceosome. Curr Opin Struct Biol 18: conservation and divergence ofHedgehog pathways from Droso- 236–242. phila to vertebrates. Development 133: 3–14. Pasca di Magliano M, Hebrok M. (2003). Hedgehog signalling in Ikram MS, Neill GW, Regl G, Eichberger T, FrischaufAM, Aberger F cancer formation and maintenance. Nat Rev Cancer 3: 903–911. et al. (2004). GLI2 is expressed in normal human epidermis and Pasca di Magliano M, Sekine S, Ermilov A, Ferris J, Dlugosz AA, BCC and induces GLI1 expression by binding to its promoter. Hebrok M. (2006). Hedgehog/Ras interactions regulate early stages J Invest Dermatol 122: 1503–1509. ofpancreatic cancer. Genes Dev 20: 3161–3173.

Oncogene Cooperative target gene regulation by GLI and JUN S Laner-Plamberger et al 1651 Pearse II RV, Collier LS, Scott MP, Tabin CJ. (1999). Vertebrate Sibilia M, Fleischmann A, Behrens A, Stingl L, Carroll J, Watt FM homologs of Drosophila suppressor offusedinteract with the gli et al. (2000). The EGF receptor provides an essential survival signal family of transcriptional regulators. Dev Biol 212: 323–336. for SOS-dependent skin tumor development. Cell 102: 211–220. Regl G, Kasper M, Schnidar H, Eichberger T, Neill GW, Ikram MS Smeal T, Binetruy B, Mercola D, Grover-Bardwick A, Heidecker G, et al. (2004a). The zinc-finger transcription factor GLI2 antagonizes Rapp UR et al. (1992). Oncoprotein-mediated signalling cascade contact inhibition and differentiation of human epidermal cells. stimulates c-Jun activity by phosphorylation ofserines 63 and 73. Oncogene 23: 1263–1274. Mol Cell Biol 12: 3507–3513. Regl G, Kasper M, Schnidar H, Eichberger T, Neill GW, Philpott MP Speek M, Njunkova O, Pata I, Valdre E, Kogerman P. (2006). A et al. (2004b). Activation ofthe BCL2 promoter in response to potential role ofalternative splicing in the regulation ofthe Hedgehog/GLI signal transduction is predominantly mediated by transcriptional activity ofhuman GLI2 in gonadal tissues. BMC GLI2. Cancer Res 64: 7724–7731. Mol Biol 7: 13. Regl G, Neill GW, Eichberger T, Kasper M, Ikram MS, Koller J Stecca B, Mas C, Clement V, Zbinden M, Correa R, Piguet V et al. et al. (2002). Human GLI2 and GLI1 are part ofa positive (2007). Melanomas require HEDGEHOG-GLI signaling regulated feedback mechanism in basal cell carcinoma. Oncogene 21: by interactions between GLI1 and the RAS-MEK/AKT pathways. 5529–5539. Proc Natl Acad Sci USA 104: 5895–5900. Riobo NA, Haines GM, Emerson Jr CP. (2006a). Protein kinase Stecca B, Ruiz i Altaba A. (2002). The therapeutic potential of C-delta and mitogen-activated protein/extracellular signal-regulated modulators ofthe Hedgehog-Gli signaling pathway. J Biol 1:9. kinase-1 control GLI activation in hedgehog signaling. Cancer Res Stein B, Angel P, van Dam H, Ponta H, Herrlich P, van der Eb A et al. 66: 839–845. (1992). Ultraviolet-radiation induced c-jun gene transcription: two Riobo NA, Lu K, Ai X, Haines GM, Emerson Jr CP. (2006b). AP-1 like binding sites mediate the response. Photochem Photobiol Phosphoinositide 3-kinase and Akt are essential for Sonic Hedgehog 55: 409–415. signaling. Proc Natl Acad Sci USA 103: 4505–4510. Toualbi K, Guller MC, Mauriz JL, Labalette C, Buendia MA, Roessler E, Ermilov AN, Grange DK, Wang A, Grachtchouk M, Mauviel A et al. (2007). Physical and functional cooperation Dlugosz AA et al. (2005). A previously unidentified amino-terminal between AP-1 and beta-catenin for the regulation of TCF-dependent domain regulates transcriptional activity ofwild-type and disease- genes. Oncogene 26: 3492–3502. associated human GLI2. Hum Mol Genet 14: 2181–2188. Vezina CM, Bushman AW. (2007). Hedgehog signaling in prostate Roth FP, Hughes JD, Estep PW, Church GM. (1998). Finding DNA growth and benign prostate hyperplasia. Curr Urol Rep 8: 275–280. regulatory motifs within unaligned noncoding sequences clustered Weston CR, Davis RJ. (2007). The JNK signal transduction pathway. by whole-genome mRNA quantitation. Nat Biotechnol 16: 939–945. Curr Opin Cell Biol 19: 142–149. Rubin LL, de Sauvage FJ. (2006). Targeting the Hedgehog pathway in Xie J, Murone M, Luoh SM, Ryan A, Gu Q, Zhang C et al. (1998). cancer. Nat Rev Drug Discov 5: 1026–1033. Activating smoothened mutations in sporadic basal-cell carcinoma. Rubinson DA, Dillon CP, Kwiatkowski AV, Sievers C, Yang L, Nature 391: 90–92. Kopinja J et al. (2003). A lentivirus-based system to functionally Yates S, Rayner TE. (2002). Transcription factor activation in silence genes in primary mammalian cells, stem cells and transgenic response to cutaneous injury: role ofAP-1 in reepithelialization. mice by RNA interference. Nat Genet 33: 401–406. Wound Repair Regen 10: 5–15. Ruiz i Altaba A, Mas C, Stecca B. (2007). The Gli code: an Young MR, Li JJ, Rincon M, Flavell RA, Sathyanarayana BK, information nexus regulating cell fate, stemness and cancer. Trends Hunziker R et al. (1999). Transgenic mice demonstrate AP-1 Cell Biol 17: 438–447. (activator protein-1) transactivation is required for tumor promo- Sasaki H, Nishizaki Y, Hui C, Nakafuku M, Kondoh H. (1999). tion. Proc Natl Acad Sci USA 96: 9827–9832. Regulation ofGli2 and Gli3 activities by an amino-terminal Zenz R, Scheuch H, Martin P, Frank C, Eferl R, Kenner L et al. repression domain: implication ofGli2 and Gli3 as primary (2003). c-Jun regulates eyelid closure and skin tumor development mediators ofShh signaling. Development 126: 3915–3924. through EGFR signaling. Dev Cell 4: 879–889. Schaefer TS, Sanders LK, Nathans D. (1995). Cooperative transcrip- Zenz R, Wagner EF. (2006). Jun signalling in the epidermis: from tional activity ofJun and Stat3 beta, a short formofStat3. Proc Natl developmental defects to psoriasis and skin tumors. Int J Biochem Acad Sci USA 92: 9097–9101. Cell Biol 38: 1043–1049. Schutte J, Minna JD, Birrer MJ. (1989). Deregulated expression of Zhang G, Luo X, Sumithran E, Pua VS, Barnetson RS, Halliday GM human c-jun transforms primary rat embryo cells in cooperation et al. (2006). Squamous cell carcinoma growth in mice and in culture with an activated c-Ha-ras gene and transforms rat-1a cells as a is regulated by c-Jun and its control ofmatrix metalloproteinase-2 single gene. Proc Natl Acad Sci USA 86: 2257–2261. and -9 expression. Oncogene 25: 7260–7266. Sheng H, Goich S, Wang A, Grachtchouk M, Lowe L, Mo R et al. Zhao M, Qiao M, Harris SE, Chen D, Oyajobi BO, Mundy GR. (2002). Dissecting the oncogenic potential ofGli2: deletion ofan (2006). The zinc finger transcription factor Gli2 mediates bone NH(2)-terminal fragment alters skin tumor phenotype. Cancer Res morphogenetic protein 2 expression in osteoblasts in response to 62: 5308–5316. hedgehog signaling. Mol Cell Biol 26: 6197–6208.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene