Glioma-Associated Oncogene 1 (GLI1)
Historical Background and GLI Family Tree
Vogelstein and colleagues discovered GLI1 (Kinzler et al. 1987) as an amplified gene in a human glioma cell line derived from a 46-year-old male patient. The investigation utilized a denaturation-renaturation gel technique to identify DNA regions that are extensively amplified which led to the observation of a greater than 50-fold amplification of a novel gene, originally named GLI (Kinzler et al. 1987). Cloning and sequencing of the GLI gene showed that its predicted protein product contained five C2H2 zinc fingers and was structurally related to the Kruppel gene family (Kinzler et al. 1988; Ruppert et al. 1988). After the cloning of additional GLI-related genes, GLI was renamed GLI1 (Ruppert et al. 1988; Ruiz i Altaba 1999).
In vertebrates, the GLI gene family consists of GLI1, GLI2, and GLI3. The encoded GLI proteins have regions of high homology, but each protein performs distinct functions in vivo (Hui and Angers 2011). GLI1 and GLI2 are primarily involved in transcriptional activation whereas GLI3 acts mainly as a repressor. However, depending on the context, GLI2 can also have repressive activity and GLI3 can behave as an activator (Aza-Blanc et al. 2000; Lo Re et al. 2012). The conversion of GLI2 and GLI3 from full-length activators to N-terminal repressor forms is accomplished by removal of the C-terminal region by proteolysis at specific sites within these proteins. In contrast, GLI1 lacks an N-terminal repressor region and is not subject to controlled C-terminal proteolysis, thus functioning only as an activator (Lee et al. 2016; Pandolfi and Stecca 2015). Two alternatively spliced GLI1 isoforms, GLI1ΔN and tGLI1, have been discovered (Lo et al. 2009; Shimokawa et al. 2008). GLI1ΔN, which lacks amino acids 1–128 of GLI1, is expressed more highly in normal tissues than in tumors, and has lower transcriptional activity than full-length GLI1 (Shimokawa et al. 2008). tGLI1, which lacks amino acids 34–74 of GLI1, is more expressed in glioblastoma and breast cancer cells than in normal cells, and stimulates cell migration and invasiveness to a greater extent than full-length GLI1 (Lo et al. 2009). Thus, alternatively spliced GLI1 isoforms increase the versatility with which GLI1 regulates transcription.
Canonical and Noncanonical GLI1 Activation
GLI1 is involved in a number of important cellular processes, especially as part of the canonical Hedgehog (HH) pathway, a signaling pathway essential for embryogenesis and adult homeostasis. The HH pathway was initially discovered in Drosophila, where much has been learned about the regulation of this signaling network (Lee et al. 2016). Drosophila possesses only one GLI analogue, Cubitus interruptus (Ci), which resembles GLI2 or GLI3, since it has an N-terminal repressor domain.
Regulation of GLI1 expression and activity can also occur independently of HH signaling (see Fig. 1). A common regulator of the GLI1 pathway is transforming growth factor beta (TGFβ). There are numerous processes in which the HH and TGFβ pathways overlap during embryonic development and cancer (Lauth and Toftgard 2007; Fernandez-Zapico 2008). The discovery that TGFβ activates the expression and activity of GLI1 was made by Dennler et al. (2007). This group further showed that GLI1 activation by TGFβ required functional SMAD3 signaling and that GLI2 is first activated leading to the induction of GLI1. Further investigations of GLI-TGFβ signaling have shown that the activation of the pathway induces the expression of BCL2 via a GLI1-SMAD4 complex (Nye et al. 2014).
The RAS family of small GTPases (KRAS, NRAS, HRAS) are important regulators of GLI1 in oncogenesis. RAS proteins signal via a kinase cascade and are able to stimulate the expression and activity of GLI1 in a number of systems (Pandolfi and Stecca 2015; Ji et al. 2007). Using transgenic mouse models, Mills and colleagues were able to demonstrate GLI1 as a key mediator of KRAS-induced pancreatic cancer (Mills et al. 2013). GLI1 can also be activated by signaling (Nakamura et al. 2013) through the MTOR/S6K1 pathway and c-MYC (Aberger and Ruiz 2014).
Biochemical Properties and Regulation
Human GLI1 is a modular 1106 amino acid protein. Key features of the protein include an N-terminal SUFU binding domain, the zinc finger region (amino acids 235–387), nuclear localization and export signals, and a C-terminal transcriptional activation domain (see Fig. 2). GLI1 targets genes for transcriptional activation by binding to the DNA base pair consensus sequence (5′-GACCACCCA-3′) in gene promoters (Kinzler and Vogelstein 1990). This target is also recognized by GLI2 and GLI3 (Hallikas et al. 2006). Recognition of GLI consensus DNA sequences is accomplished by the GLI1 zinc finger domains, of which zinc fingers 4 and 5 play the most significant role (Infante et al. 2015; Pavletich and Pabo 1993). Numerous genes regulated by GLI proteins have been identified, including genes involved in development, proliferation, stemness, and transformation (Pandolfi and Stecca 2015; Aberger and Ruiz 2014; Harris et al. 2011; Palle et al. 2015).
The role of SUFU in negative regulation of GLI1 was mentioned above. Several studies have contributed to a mechanistic understanding of how SUFU inhibits GLI1 signaling. First, SUFU was found to compete with the nuclear import proteins, importins, for mutually exclusive binding to the GLI1 N-terminal SUFU-binding region (Shi et al. 2014; Szczepny et al. 2014). Thus, the binding of SUFU prevents GLI1 translocation to the nucleus by importins. Second, SUFU is sometimes transported to the nucleus with GLI1, where it forms a complex with SAP18, SIN3, and histone deacetylases (Hui and Angers 2011). When this complex binds with GLI1 at target gene promoters, transcription of these genes is repressed.
Mechanistically, GLI1 is thought to activate gene expression by recruiting chromatin modifying enzymes to specific gene promoters leading to regulation of transcription. Several GLI1-binding chromatin cofactors have been identified. For example, GLI1 has been demonstrated to physically interact with the histone acetyltransferase, p300/CBP-associated factor (PCAF) (Nye et al. 2014; Mazza et al. 2013). In these studies, PCAF was found to be required for GLI1-dependent transcriptional activation and to positively regulate GLI1-dependent transcription when overexpressed. Notably, PCAF required GLI1 for it to be targeted to GLI-binding regions of GLI target promoter, and the presence of PCAF at these promoters was associated with increased histone 3 lysine acetylation (Nye et al. 2014; Mazza et al. 2013). These results suggest that PCAF is recruited to GLI1 target promoters, where it acetylates histones, thus facilitating gene transcription.
GLI1 has also been shown to interact with member proteins of SWI/SNF (SWItch/sucrose nonfermentable) nucleosome remodeling complexes. This large complex is composed of many proteins, including an ATPase, either SMARCA2 (Brahma) or SMARCA4 (BRG1). Jagani et al. demonstrated that GLI1 overexpressed in mouse TM3 Leydig cells was associated with SNF5 and two other SWI/SNF complex members (Jagani et al. 2010). Furthermore, SNF5 was localized to promoters of GLI target genes, and knockdown of SNF5 caused a decrease in the expression of GLI targets in TM3 cells. A second set of studies demonstrated that BRG1 was responsible for repression of GLI targets in the unstimulated state via interactions with GLI3. However, HH-induced increases in expression of GLI target genes required BRG1 interaction with GLI1 (Shi et al. 2016; Zhan et al. 2011). The ability of BRG1 to support HH signaling did not require BRG1 ATPase activities (Zhan et al. 2011), suggesting that BRG1 might act via a different mechanism besides nucleosome remodeling. BRG1 was localized to the promoters of a number of GLI target genes, including GLI1, PTCH1, and NKX6.1, at or near regions bound by GLI1 in developing mouse telencephalon (Zhan et al. 2011). Thus, SWI/SNF complexes may be involved in both the positive and negative regulation of HH/GLI signaling.
Another GLI1 cofactor is parafibromin (gene name HRPT2), a member of the PAF1 complex, which is involved in transcriptional initiation and elongation. Depletion of parafibromin inhibits GLI1-dependent transcription and GLI1 was demonstrated to bind directly to parafibromin (Mosimann et al. 2009). Parafibromin bound to an N-terminal region of GLI3 overlapping the SUFU-binding region, and thus it was suggested that parafibromin binds GLI1 at this region of GLI1 as well (Mosimann et al. 2009). Recently, it was demonstrated that GLI1 selectively binds the dephosphorylated form of parafibromin and that dephosphorylated parafibromin increases GLI1 transcriptional activity (Kikuchi et al. 2016). In addition, GLI1 competes with the WNT pathway effector, β-catenin, for binding to dephosphorylated parafibromin. Thus, greater expression of activated β-catenin reduces the binding of GLI1 to parafibromin, reducing GLI1 activity, and vice versa (Kikuchi et al. 2016). Therefore, parafibromin has the potential for integrating signals from the HH and WNT pathways.
GLI1 also binds the TFIID transcription factor complex subunit, TAF9 (Yoon et al. 2015; Yoon et al. 1998). An 18-amino acid region (amino acids 1020–1091) within the transcriptional activation domain of GLI1 (see Fig. 2) was found to be highly homologous to the α-helical herpes simplex viral protein 16 activation domain, suggesting that GLI1 binds to TAF9 (Yoon et al. 1998). GLI1 and GLI2 were shown to bind TAF9 by coimmunoprecipitation in cells and directly by in vitro methods, and to positively regulate GLI1-driven transcription (Yoon et al. 2015; Bosco-Clement et al. 2014). In the presence of wild-type p53, which binds TAF9 more strongly than GLI1, TAF9 interaction with GLI proteins is reduced leading to decreased GLI1 transcription (Yoon et al. 2015). This observation provides one mechanism by which p53 inhibits GLI1 expression (Stecca and Ruiz i Altaba 2009).
In addition to GLI1 regulation by cooperation with transcription cofactors, GLI1 is extensively post-translationally modified, which plays multiple roles in its localization, stability, and activity. Protein kinase A (PKA) has been shown to phosphorylate GLI1 at Thr374, a site near a GLI1 nuclear localization signal (Sheng et al. 2006). Phosphorylation of this site decreases the targeting of GLI1 to the nucleus, thus reducing GLI1 transcriptional activity (Sheng et al. 2006). In contrast, phosphorylation of GLI1 by DYRK1A and S6K1 promotes GLI1 nuclear translocation and activity (Aberger and Ruiz 2014; Mao et al. 2002). aPKC-ι/λ phosphorylates GLI1 within its zinc finger region at Ser243 and Thr304, which increases GLI1 binding to DNA without affecting GLI1 nuclear localization (Atwood et al. 2013). GLI1 degradation is regulated by its binding to the adaptor protein, NUMB, which activates the ubiquitin E3 ligase, ITCH. Ubiquitination of GLI1 marks the protein for proteosomal degradation (Di Marcotullio et al. 2011). Ubiquitination and proteosomal degradation of GLI1 are promoted by adenine monophosphate kinase (AMPK), an enzyme activated by low cellular ATP, which phosphorylates GLI1 at Ser102 and Ser408 and Thr1074 (Li et al. 2015; Xu et al. 2014) and opposed by GLI1 phosphorylation by IKKβ (Agarwal et al. 2016). Under conditions of genotoxic stress, GLI1 is also ubiquitinated by PCAF, resulting in GLI1 degradation (Mazza et al. 2013). In contrast to ubiquitination, modification of GLI1 by SUMOlation at Lys180 and Lys815 results in increased stability of the protein (Cox et al. 2010).
Reports of interactions between GLI1 and other proteins suggest additional functions for GLI1. GLI1 has been demonstrated to interact with p-TEFb, a positive transcription elongation factor that functions as part of the RNA polymerase II pausing complex, and with the DNA replication licensing factors CDT1, ORC4, and MCM2 (Zhang et al. 2016). In addition, the deubiquitinase, USP21, binds to and stabilizes GLI1 at the centrosome (Heride et al. 2016). These protein interactions are not fully characterized but suggest the involvement of GLI1 with RNA polymerase complexes and cellular replication processes.
GLI1 in Development and Differentiation
The role of GLI1 in development and HH signaling was explored through the generation of GLI1 and GLI2 mutant transgenic mice (Park et al. 2000). Homozygous GLI2 or GLI3 mutant mice developed multiple defects, whereas homozygous GLI1 mutant mice had no apparent deficiencies. However, while mice expressing only one mutant allele for GLI2 were normal, mice which were homozygous for the GLI1 mutation and heterozygous for the GLI2 mutation exhibited defects similar to GLI2 homozygous mutant mice. This study suggested that the function of GLI1 is partially redundant with that of GLI2. A second study by the same group using GLI1 null mice had similar findings (Bai et al. 2002). These studies further suggested that in development, GLI1 is only expressed in response to HH ligand and that initial HH signaling is largely accomplished by basally expressed GLI2. The importance of GLI1 was further evaluated using GLI1 knockout mice in the context of hematopoiesis (Merchant et al. 2010). Their results showed that the loss of GLI1 caused the impairment of the normal stress response of hematopoietic stem cells (HSCs), along with the decreased proliferation and differentiation of myeloid progenitors. Further investigation showed that GLI1 null HSCs had diminished expression of the GLI1 target, Cyclin D1, suggesting a mechanism for the decreased proliferation observed in GLI1 null HSCs (Merchant et al. 2010). Increasing evidence has demonstrated that GLI1 plays key roles in stem cells in various contexts (Denham et al. 2010; Kramann et al. 2015a). Thus, while GLI1 may have overlapping functions with GLI2 during development, it may play significant nonredundant roles in specific differentiation and regeneration pathways.
GLI1 in Cancer
GLI1 is sometimes overexpressed in various types of cancer due to germline or somatic mutations affecting genes in the SHH-Gli pathway, such as constitutively activating SMO, or deactivating PTCH1 or SUFU mutations in basal cell carcinomas (BCC), medulloblastomas, and rhabdomyosarcomas (Di Magno et al. 2015; Rimkus et al. 2016). As mentioned above, GLI1 expression is elevated in many malignancies (e.g., cancers of the brain, skin, pancreas, prostate, and breast) due to the activation of oncogenic pathways such as RAS, TGFβ, or WNT (Pandolfi and Stecca 2015; Palle et al. 2015; Amakye et al. 2013). GLI1 regulates the expression of many genes that have the potential to impact significant malignant characteristics of human cancers such as the ability to proliferate, migrate, and invade. The capability to divide and proliferate without restriction is a principal feature of cancer cells. GLI1 targets include the transcriptional factor FOXM1, IGFBP6, PDGFRα, and Cyclin D1/D2, all of which can modulate cell cycle progression and thus proliferation (Pandolfi and Stecca 2015). Cell migration and invasion is promoted by GLI1-induced genes such as osteopontin (Harris et al. 2011). Osteopontin protein along with GLI1 is highly expressed in melanomas, and its expression has been linked to the ability of melanomas to migrate, invade, and metastasize. The capacity to invade and metastasize is often accompanied by epithelial mesenchymal transition (EMT) of cancer cells. GLI1 mediates this dedifferentiation program by increasing the expression of transcription factors such as SNAI1, SNAI2, ZEB1, and ZEB2, which repress expression of E-cadherin, a major determinant of epithelial tissue organization and cell polarity (Pandolfi and Stecca 2015; Katoh and Katoh 2009). In addition, GLI1 can regulate the expression of antiapoptotic factors (BCL2 and BFL1/A1) and drug transporters (ABCG2 and ABCB1) to promote survival in cancer cells (Chen et al. 2014; Comba et al. 2016). Tumor-initiating cells, also known as cancer stem cells, are proposed to arise from existing cancer cell populations or alternatively from the transformation of normal tissue stem cells. These cells are important for the self-renewal and generation of cells that form a tumor. GLI1 is often overexpressed and has potent proproliferative effects on cancer stem cells (Stecca and Ruiz 2010; Aberger and Ruiz 2014; Harris et al. 2011). It is also important to recognize that GLI1 can have either a tumor suppressor or a tumor promoter function depending upon cellular context and mutational profile. For example, knockout of GLI1 decreased KRAS mutant-driven pancreatic neoplasia in mice in the presence of endogenous p53, but accelerated pancreatic cancer in Kras mutant mice in which p53 was also deleted (Mills et al. 2013, 2014).
In addition to the involvement of GLI1 in oncogenicity by expression in cancer cells, GLI1 plays important roles in the biology of stromal cells (e.g., fibroblasts, endothelial cells, and immune cells) that surround various types of tumor. Many cancer cells produce HH, increasing GLI1 in the neighboring stromal cells via paracrine HH/GLI1 signaling in a canonical manner, thus influencing the tumor microenvironment (Stecca and Ruiz 2010; Merchant and Saqui-Salces 2014). A major component of the tumor stroma is fibroblasts. These cells secrete extracellular matrix proteins and growth factors that can contribute to tumor growth. Pancreatic fibroblasts stimulated with HH in vitro increased GLI1 expression and exhibited increased proliferation, consistent with the idea that GLI1 regulates cell growth in stromal cells (Hwang et al. 2012). A demonstration of a role for GLI1 in cancer-associated immune cells is provided in the case of Helicobacter-induced metaplasia. Helicobacter infection of the stomach increases inflammation leading to increased incidence of gastric cancer. Upon infection, gastric epithelial cells secrete HH which recruits immune cells to the stomach (Schumacher et al. 2012). However, in mice with deletion of GLI1, after 6-month infection with Helicobacter, metaplasia did not develop and certain classes of myeloid cells were absent in stomach samples and cytokine composition was altered (El-Zaatari et al. 2013).
The downregulation of HH pathway components, such as GLI and SMO, is being explored for its therapeutic potential for a number of different cancers and diseases. The first drug shown to inhibit GLI1 expression by targeting Smo in the canonical pathway is the natural product cyclopamine. Unfortunately, the usage of this alkaloid is limited due to its toxicity and teratogenicity. The FDA has also approved the Smo inhibitors, vismodegib and sonidegib, for the treatment of basal cell carcinoma, and additional Smo antagonists are under development (Rimkus et al. 2016). Like cyclopamine, they have many unwanted adverse effects, especially in pregnant women. An alternative SMO antagonist is the systemic antifungal agent itraconazole. This antagonist uses different mechanisms to vismodegib and other SMO antagonists and can be used in combination with cyclopamine for more effective tumor inhibition (Kim et al. 2010).
Screening for direct inhibitors of GLI1 led to the development of GANT61 (Gli antagonist 61). This GLI1 antagonist led to increased apoptosis in tumor cells and blocked GLI1 and GLI2 transcriptional activity more effectively than cyclopamine (Lauth et al. 2007). GANT61 is reported to bind the region between zinc fingers 1 and 2 of GLI1 and to inhibit the interaction of GLI1 with DNA, but does not interact directly with the DNA binding region of GLI1 (Agyeman et al. 2014). GANT61 is presumed to act similarly upon GLI2. The natural product Glabrescione B is another GLI1 antagonist that inhibits tumor cell growth, self-restoration, and clonogenicity of tumor-derived stem cells. It has been proven to be effective against medulloblastomas as well as basal cell carcinomas by interfering with DNA (Infante et al. 2015). Glabrescione B was shown to bind GLI1 between zinc fingers 4 and 5 and directly inhibits the binding of GLI1 to DNA (Infante et al. 2015). In addition, arsenic trioxide (ATO) is also a GLI1/GLI2 antagonist that is an FDA approved choice of treatment for acute promyelocytic leukemia (Agyeman et al. 2014). ATO is reported to bind directly to GLI1 but does not inhibit GLI1/DNA interaction (Beauchamp et al. 2011). Another arsenical compound that also inhibits GLI1 and GLI2, darinaparsin, is in clinical trials. Darinaparsin is proposed to cause increased stabilization of GLI2, leading to the loss of GLI1 expression (Kramann et al. 2015b). The development of compounds that directly target GLI proteins increases the possibilities for therapeutic approaches in cancers in which GLI supports tumorigenesis.
GLI1 and Nonneoplastic Diseases
A number of human diseases including birth defects and cancers are associated with abnormal functions of the HH/GLI signaling pathway. GLI1 itself is rarely mutated, although GLI1 germline variants have been reported in inflammatory bowel disease and Hirschsprung disease (Palle et al. 2015). Inherited mutations in GLI2 and GLI3 result in congenital craniofacial and limb development defects such as holoproencephaly and Greig cephalopolysyndactyly syndromes, respectively (Cohen 2010). GLI1 expression during development is undoubtedly altered in these diseases as a result of GLI2 or GLI3 disruption. One study explored the significance of GLI1 for salivary gland development using transgenic mice in which GLI1 was overexpressed in salivary epithelial cells (Fiaschi et al. 2011). GLI1 overexpression caused decreased epithelial cell differentiation but increased cell proliferation, with increased expression of the GLI1 target, Cyclin D1. Morphologically, an expansion of salivary ductal structures and a concomitant loss of acini were observed. This study illustrates how the disturbance of GLI1 expression can lead to developmental alterations.
The HH-GLI pathway has also been implicated in organ fibrosis, such as in the liver or kidney. Organ fibrosis develops after tissue injury and entails the proliferation of fibroblast-like cells (myofibroblasts) that secrete extensive extracellular matrix proteins. This process is very similar to the proliferation of fibroblasts that occurs in many types of tumor. Chronic injury to the tissues often induces HH expression, which acts in a paracrine fashion increasing GLI1 expression in nearby fibroblasts (Bolanos et al. 2012; Choi et al. 2011). GLI1 in these cell types could also be increased by noncanonical pathways (e.g., TGFβ). Recent studies suggest that many myofibroblasts derive from GLI1-expressing mesenchymal stem cells, and that GLI1 expression and fibrogenic behavior in these cells is dependent upon GLI2 (Kramann 2016). Thus, the processes of fibroblast activation in fibrosis and tumorigenesis appear to be closely linked to the roles of GLI in mesenchymal stem cells.
It is clear that GLI1 and hedgehog play a significant role in development, differentiation, and cancer through their ability to transcriptionally activate a number of key gene targets. GLI1 is a central signaling molecule not only in the HH pathway but also in a growing number of additional signaling networks. The expression of GLI1 is tightly controlled in normal cells, and the ability of the GLI1 protein to interact with other proteins and be extensively post-translationally modified lends great flexibility to the regulation of its activity and stability. GLI1 is frequently overexpressed in cancer, where it plays roles in both tumor cells and surrounding stromal cells. Studies have shown that GLI1 may be an appropriate therapeutic target for cancer treatment. However, further research is needed to fully comprehend the function and influence of GLI1 in cancerous tissue, especially in terms of the cross-talk and paracrine signaling networks within the tumor microenvironment.
- Bolanos AL, Milla CM, Lira JC, Ramirez R, Checa M, Barrera L, et al. Role of Sonic Hedgehog in idiopathic pulmonary fibrosis. Am J Phys Lung Cell Mol Phys. 2012;303(11):L978–90.Google Scholar
- Comba A, Almada LL, Tolosa EJ, Iguchi E, Marks DL, Vara Messler M, et al. Nuclear factor of activated T cells-dependent down-regulation of the transcription factor glioma-associated protein 1 (GLI1) underlies the growth inhibitory properties of arachidonic acid. J Biol Chem. 2016;291(4):1933–47.PubMedCrossRefGoogle Scholar
- Lo Re AE, Fernandez-Barrena MG, Almada LL, Mills L, Elsawa SF, Lund G, et al. A novel AKT1-GLI3-VMP1 pathway mediates KRAS-induced autophagy in cancer cells. J Biol Chem. 2012;287(30):25325–34Google Scholar
- Mills LD, Zhang Y, Marler RJ, Herreros-Villanueva M, Zhang L, Almada LL, et al. Loss of the transcription factor GLI1 identifies a signaling network in the tumor microenvironment mediating KRAS oncogene-induced transformation. J Biol Chem. 2013;288(17):11786–94.PubMedCrossRefPubMedCentralGoogle Scholar
- Nye MD, Almada LL, Fernandez-Barrena MG, Marks DL, Elsawa SF, Vrabel A, et al. The transcription factor GLI1 interacts with SMAD proteins to modulate transforming growth factor beta-induced gene expression in a p300/CREB-binding protein-associated factor (PCAF)-dependent manner. J Biol Chem. 2014;289(22):15495–506.PubMedCrossRefPubMedCentralGoogle Scholar