Targeting the Wnt Pathway
Writer and Curator: Larry H Bernstein, MD, FCAP
7.11 Targeting the Wnt Pathway
7.11.1 Targeting the Wnt pathway in human cancers. Therapeutic targeting with a focus on OMP-54F28
7.11.2 Wnt signaling and hepatocarcinogenesis – Molecular targets
7.11.4 SALL4 is directly activated by TCF.LEF in the canonical Wnt signaling pathway
7.11.5 SALL4. An emerging cancer biomarker and target
7.11.6 Sal-like 4 (SALL4) suppresses CDH1 expression and maintains cell dispersion in basal-like breast cancer
7.11.7 The transcription factor SALL4 regulates stemness of EpCAM-positive hepatocellular carcinoma
7.11.8 Overexpression of the novel oncogene SALL4 and activation of the Wnt.β-catenin pathway in myelodysplastic syndromes
7.11.1 Targeting the Wnt pathway in human cancers. Therapeutic targeting with a focus on OMP-54F28
Le PN, McDermott JD, Jimeno A.
Pharmacol Ther. 2015 Feb; 146:1-11
The Wnt signaling pathways are a group of signal transduction pathways that play an important role in cell fate specification, cell proliferation and cell migration. Aberrant signaling in these pathways has been implicated in the development and progression of multiple cancers by allowing increased proliferation, angiogenesis, survival and metastasis. Activation of the Wnt pathway also contributes to the tumorigenicity of cancer stem cells (CSCs). Therefore, inhibiting this pathway has been a recent focus of cancer research with multiple targetable candidates in development. OMP-54F28 is a fusion protein that combines the cysteine-rich domain of frizzled family receptor 8 (Fzd8) with the immunoglobulin Fc domain that competes with the native Fzd8 receptor for its ligands and antagonizes Wnt signaling. Preclinical models with OMP-54F28 have shown reduced tumor growth and decreased CSC frequency as a single agent and in combination with other chemotherapeutic agents. Due to these findings, a phase 1a study is nearing completion with OMP-54F28 in advanced solid tumors and 3 phase 1b studies have been opened with OMP-54F28 in combination with standard-of-care chemotherapy backbones in ovarian, pancreatic and hepatocellular cancers. This article will review the Wnt signaling pathway, preclinical data on OMP-54F28 and other Wnt pathway inhibitors and ongoing clinical trials.
Wnt signaling pathway
Three Wnt signaling pathways have been deﬁned, including the canonical, non-canonical planar cell polarity pathway and the noncanonical Wnt/Ca2+ pathway.Of the three,the canonicalWnt pathway is the best described. Here, a cysteine-rich Wnt ligand binds the extracellular cysteine-rich domain (CRD) at the amino terminus of a seven pass transmembrane receptor termed Frizzled (FZ/Fzd [Vinson et al., 1989; Bhanot et al., 1996]) and low-density lipoprotein (LDL) receptor-related protein 5/6 (LRP5/6) that acts as a co-receptor (Pinson et al., 2000; Tamai et al., 2000; Wehrli et al., 2000) to start the activation of the canonical Wnt signaling pathway. Nineteen Wnt ligands have been identiﬁed along with 10 Fzd receptors (Huang & Klein,2004).Various Wnt ligands have been shown to bind to particular Fzd receptors, but this interaction is promiscuous wherein oneWnt can bind multiple Fzd receptors (Bhanot et al., 1996). Wnt glycoproteins are relatively hydrophobic and insoluble possibly due to cysteine palmitoylation by Porcupine(PORC [Willertetal., 2003; Zhai et al., 2004]). However, PORC is required for Wnt signaling, suggesting that palmitoylation is essential in Wnt ligand secretion and pathway activation. Wnt ligands can activate signaling by both autocrine and paracrine signaling (Baﬁco et al., 2004). Wnt signaling can be inhibited through the binding of soluble Dickkopf (DKK) to LRP5/6 (Glinkaetal.,1998) or secreted Frizzled-related protein (SFRP) binding to Wnt ligands due to their sequence homology to the CRD domain of Fzd (Hoang et al., 1996). Wnt inhibitor factor (WIF) proteins, due to their similarity to the extracellular domain of derailed/RYK Wnt transmembrane receptors, can also regulate Wnt signaling by interacting with Wnt ligands (Hsieh et al., 1999a). When there is no Wnt ligand present, β-catenin levels are limited by the destruction complex that includes Adenomatous Polyposis Coli (APC) and AXIN. With Wnt signaling “off,” AXIN facilitates the phosphorylation of β-catenin by casein kinase 1 (CK1) and glycogen synthase kinase 3 (GSK-3 [Peifer et al., 1994; Yost et al., 1996; Sakanaka etal.,1999;Liuetal.,2002]). These phosphorylated ser/thr sites are recognized by an E3ubiquitin ligase complex, and β-catenin is subsequently targeted for proteasomal degradation (Aberleetal.,1997). Therefore, β-catenin is maintained at low cytoplasmic and nuclear levels. In the “on” state, Wnt ligand binds the extracellular CRD of the amino terminus of Fzd and the LRP5/6 co-receptor (Dannet al., 2001; Pinson et al., 2000;Tamaietal.,2000). Dishevelled (Dsh/Dvl) is activated and recruited along with the destruction complex to the plasma membrane (Lee et al., 1999;Rothbacher et al., 2000).AXIN alsointeracts with the plasma membrane, possibly by binding the cytoplasmic tail of LRP5/6 (Mao et al., 2001). This binding is promoted by phosphorylation of LRP5/6 by GSK-3 and CK1 (Davidson et al., 2005; Zeng et al., 2005). AXIN is degraded, and GSK-3 is thus prevented from phosphorylating β-catenin. This leads to the accumulation of β-catenin in the nucleus and its interaction with T-cell factor (TCF) and lymphoid enhancerbinding protein (LEF) transcription factors to activated downstream targets (Behrenset al., 1996;Huber et al., 1996).
Wnt pathway and cancer
Aberrant Wnt signaling was ﬁrst implicated in cancer in mouse studies, where mouse mammary tumor virus (MMTV) was found to be virally inserted into the promoter region of Int-1, promoting mammary tumors (Nusse & Varmus, 1982; Tsukamoto et al., 1988). It was later found that Int-1 was a homologue to Wg, and thus renamed Wnt (Nusse et al., 1991; Rijsewijk et al., 1987). Since this time, the Wnt pathway has been shown to be aberrantly regulated in many cancers. Abnormal β-catenin activation has been well characterized in colon cancer, where mutations in APC, or less frequently in β-catenin, results in constitutively active β-catenin and consequently active downstream effectors (Morin et al., 1997). While APC and β-catenin mutations are rare in lung cancer, overexpression of Dvl, Wnt-1 and Wnt-2 have all been correlated with non-small cell lung cancer (NSCLC) (He et al., 2004; Pongracz & Stockley, 2006; Ueda et al., 2001; Uematsu et al., 2003; You et al., 2004c). Moreover, increased tumor relapse was associated with a TCF4 Wnt gene signature in lung adenocarcinomas (Nguyen et al., 2009b). Together, these data provide strong evidence for the role of Wnt signaling in lung cancers. Wnt-5a has also been shown to be increased in breast cancer (Lejeune et al., 1995). Several of Fzds that have been shown to be overexpressed in cancers and/or cancer cell migration include Fzd4, Fzd7, Fzd8 and Fzd10 (Fukukawa et al., 2009; Jin et al., 2011; Ueno et al., 2009; Wang et al., 2012b; Yang et al., 2011). These have been shown to activate the canonical and/or non-canonical Wnt pathway. However, these are just a few of the studies linking Fzd overexpression with cancer, and an extensive list was previously covered by Ueno et al. (2013). Wnt expression has also beenassociated with metastasis and tumor microenvironment. Inhibition of Wnt signaling byRNAi targeting LEF1 and HOXB9 reduced brain and bone metastasis using a mouse model of lung adenocarcinoma (Nguyen et al., 2009b). The mechanism of LEF1 and HOXB9 metastasis promotion was not elucidated in this study although Wnt signaling, speciﬁcally Wnt-1 and Wnt-5a, has been shown to increase proliferation and survival of endothelial cells (Masckauchan et al., 2005,2006). β-catenin was also shown to correlate with VEGF expression in colon cancer (Easwaran et al., 2003; Zhang et al., 2001), suggesting ar ole for Wnt signaling in angiogenesis.Moreover, Wnt-5a expression has recently been shown to be increased in NSCLC; its expression in patient tissue was correlated with expression of angiogenesis related proteins such as vascular endothelial cadherin and matrix metalloprotease 2, microvessel density and vasculogenic mimicry, all of which suggest a role for Wnt-5a in promoting angiogenesis (Yao et al., 2014). Decreased expression of Wnt pathway inhibitors (WIF-1,DKKs, and SFRPs) “allows” for the activation of Wnt signaling and has also been observed in various cancers.For example, the down-regulation of associated Wnt antagonist, WIF-1, has been implicated in the breast, prostate, lung and bladder cancer (Wissmann et al., 2003). Furthermore, WIF-1 has been shown to be epigenetically silenced in lung and bladder cancer (Mazieres et al., 2004; Urakami et al., 2006). Epigenetic silencing of DKK-1 has been shown in colorectal cancer (Aguilera et al., 2006) and SFRP in NSCLC, hepatocellular carcinoma and colorectal cancer (Fukui etal., 2005; Shihetal., 2006; Suzukietal., 2004). Recent studies suggest that Wnt inhibitors may also play a pro-apoptotic role, where reduced apoptosis and p53 expression were observed in mammary glands isolated from SFRP-/- mice following induction of DNA damage by -irradiation (Gauger & Schneider, 2014). In addition, another study suggests that WIF- 1 may inhibit angiogenesis. DKK-1 and WIF-1 directly interact and together may act as co-regulators in promoting apoptosis in the human umbilical vein endothelial cell (HUVEC) system(Koetal.,2014). Although Wnt signaling is not as well correlated with head and neck squamous cell carcinoma (HNSCC) as with other cancers, such as colon cancer, recent studies provide evidence that Wnt signaling is an attractive target in HNSCC. Wnt pathway activation has been shown in HPV positive HNSCC, possibly driven by E6 and E7 (Rampias et al., 2010). β-catenin nuclear accumulation was also observed in the majority of patient HNSCC tumor samples (Wend et al., 2013). Up-regulation of several Fzd receptors was observed in HNSCC, including Fzd1, Fzd7a, Fzd10b, Fzd2 and Fzd13 (Rhee et al., 2002). Furthermore, Wnt expression may affect radio sensitivity in HNSCC cell lines, where β-catenin nuclear accumulation was correlated with radiation-resistance (Chang et al.,2008). Similarly, radiation-resistant mouse mammary progenitor cells were associated with active Wnt signaling (Chen et al., 2007; Woodward et al., 2007). Wnt expression has been correlated with therapy resistance in prostate cancer, where Wnt16B increased following therapy and lessened DNA damage following treatment with a topoisomerase inhibitor (Sun et al., 2012). In this study, Wnt16B increased growth and proliferation. Taken together, these studies suggest that Wnt expression not only promotes cancer cell proliferation, but may also affect treatment efﬁcacy. Furthermore, the up-regulation of Wnt16B originating speciﬁcally in the stroma compartment, and through tumor-stroma interactions promoting therapy resistance in the tumor compartment, suggests that the stroma is a favorable target for therapy. Consistent with this,human ovarian ﬁbroblasts released Wnt16B in to the stroma compartment following DNA damage by radiation or chemotherapy (Shen et al., 2014). Interestingly stromal Wnt16B activated the Wnt signaling pathway in dendritic cells (DCs), causing the release of interleukin-10 (IL-10) and tumor growth factor-β (TGF-β) and regulatory T-cell differentiation. Thus, Wnt16B may not only confer therapy resistance, but also alter the tumor microenvironment and as the authors suggest, possibly promote immune evasion.
Wnt signaling and cancer stem cells (CSCs)
Wnt signaling is important in stem cell homeostasis.In the intestinal villi Wnt signaling is particularly important in stem cell maintenance as well as in determining stem cell fate (Batlle et al., 2002; Korinek et al., 1998). Wnt signaling has also been shown to be essential in stem cell proliferation and hair follicle development and may function to activate stem cells in the bulge to more proliferative progenitor cells, as well as determining cell fate (Andl et al., 2002; Choi et al., 2013; Lien et al., 2014; Lowry et al., 2005).Similarly,Wnt overexpression in hematopoietic stem cells leads to the expansion of progenitor cells, suggesting that Wnt signaling is also important in hematopoiesis (Austin et al., 1997). Aberrant Wnt signaling in the stem cell compartment has been shown to contribute to tumorigenesis. Here, it is important to note thatwhile some authors appropriately choose conservative terminology in the deﬁnition of CSCs, for the purpose of coherency in this review, we loosely combine tumor-initiating, tumor propagating and CSCs into one term, as CSCs. Loss of APC, consequently leading to the accumulationof β-catenin, in colorectal cells resulted in cells maintaining a phenotype similar to progenitor cells of the crypt (Sansom et al., 2004). In another approach, high levels of Wnt expression were observed in CSCs from colon cancer grown as spheroids (Vermeulen et al., 2010). Similarly, Wnt-1, -3 and -5a all promoted murine mammosphere growth, a method that enriches for stem cells, and results suggested both canonical and non-canonical Wnt signaling could promote growth (Many & Brown, 2014). Furthermore, hair follicle tumors were observed to have stable expression of β-catenin in mice (Gat et al.,1998). A recent study found hair follicle stem cells (HFSC) treated with dimethylbenzanthracene (DMBA) and 12-O-Tetradecanoylphorbol-13-acetate (TPA) induced sebaceous neoplasms in C57BL/6 mice, as well as increased Wnt10b expression in basal cells via immunostaining (Qiu et al., 2014). Here the authors propose a model wherein increased Wnt10b results in proliferation and differentiation of HFSCs and thus promoting sebaceous neoplasms. High levels of Wnt expression were also observed in granulocyte-macrophage progenitors isolated from chronic myeloid leukemia (CML) patients and correlated with increased self-renewal (Jamieson et al., 2004). Fzd4 was suggested to regulate “stemness” of cancer cells and promote invasiveness in glioma cells (Jin et al., 2011). In HNSCC cell lines, side populations sorted by Hoechst efﬂux, a functional assay for enriching stem cells, were more invasive and tumorigenic in nude mice, and importantly these populations exhibited higher Wnt signaling (Song et al., 2010). Together, the data suggest that the same Wnt signaling mechanisms that regulate stem cells, when abnormal, may contribute to the tumorigenic potential of CSCs.
Targeting the Wnt pathway
Wnt pathway components are often difﬁcult to target due to their redundancy in other functions. β-catenin, for example, also interacts with E-cadherin, an interaction that is essential for cell adhesion, as well as interacting with APC and TCF competitively within the same armadillo repeat domain (Behrens et al., 1996; Hulsken et al., 1994; Ozawa et al., 1989). In order to circumvent this, speciﬁc inhibitors that disrupt the β-catenin and TCF interaction have been widely explored, as well as RNAi approaches. However, even with utmost speciﬁcity, due to the essential role of the Wnt pathway in stem cell maintenance, tissue homeostasis and cell fate determination, targeting this signaling pathway has potential pitfalls. A potential concern is that toxicity, speciﬁcally to the GI tract, as well as anemia and immune suppression, might be too great for obtaining an adequate therapeutic index. In spite of these potential hurdles, research toward identifying potent Wnt pathway antagonists for cancer treatment has been promising.
Non-steroidal anti-inﬂammatory drugs (NSAIDS), vitamins A and D, and polyphenols, such as curcumin and resveratrol, have all been shown to inhibit the Wnt pathway, and these have been elegantly reviewed (Table 1 and Fig. 1 [Barker & Clevers, 2006; Takahashi-Yanaga & Sasaguri, 2007; Takahashi-Yanaga & Kahn, 2010]). These compounds, although promising, have shown insufﬁcient efﬁcacy and thus may prove ineffectual as single-agent treatments. For example, the use of NSAIDS, speciﬁcally sulindac, in patients diagnosed with Familial Adenomatous Polyposis (FAP) reduced the number of polyps by only ~44% (Giardiello et al., 1993). Quercetin, a polyphenol and dietary ﬂavonoid, has also been shown to decrease β-catenin and TCF protein levels (Fig.1) and inhibit colon cancer cell growth in vitro via decreased cyclin D1 and survivin levels (Park et al., 2005; Shan et al., 2009). Quercetin was also shown to inhibit murine mammary cancer cell growth and target theWnt pathway through DKK1,2,3 and 4 up-regulation (Kimetal., 2013). Salinomycin, an antibacterial potassium ionophore, was ﬁrst identiﬁed by high throughput screening and was shown to inhibit breast CSCs (Gupta et al., 2009). Its mechanism was later elucidated and was shown to inhibit LRP5/6 phosphorylation, causing its degradation (Fig. 1 [Lu et al., 2011a]). Salinomycin has recently been shown to inhibit breastand prostate cancer cell proliferation and induce apoptosis, targeting Wnt signaling by decreased LRP5/6 expression, but also by targeting mTORC (Lu & Li, 2014), suggesting it may function in targeting multiple pathways. Salinomycin has also been shown to have antitumorigenic effects in hepatocellular carcinoma, osteosarcoma, gastric cancer, NSCLC and nasopharygeal carcinoma; studies suggest that it speciﬁcally targets CSCs by inhibiting cell proliferation, inducing apoptosis and limiting cell migration (Arafat et al., 2013; Mao et al., 2014; Tang et al., 2011; Wang et al., 2012a; Wu et al., 2014). COX-2 inhibitors may target the Wnt pathway by inhibiting prostaglandin E2 (PGE2), the product of COX-2, which acts to phosphorylate GSK-3 (Fig. 1 [Fujino et al., 2002]). Celecoxib, a NSAID and a COX-2 inhibitor, has been shown to decrease CD133 expression, a surface marker of prostate CSCs, by targeting the Wnt pathway, and this effect was observed to be independent of its COX-2 inhibiting activity (Deng et al., 2013). In order to circumvent the toxicities associated with long term COX-2 inhibition, one group suggests using synthetic derivatives of sulindac, another NSAID that was previously mentioned, that do not target COX-2 and were successful in limiting colon cancer cell growth and promoting apoptosis in vitro(Li et al., 2013;Whitt e tal., 2012). Resveratrol has recently been shown to inhibit the growth of breast CSCs both in invitroandwhenimplantedinNOD/SCIDmicebytargetingthecanonicalWntpathwayandinducingautophagy(Fuetal.,2014).Resveratrol also limited growth of cervical cancer cells by causing cell cycle arrest and inducing apoptosis (Zhang et al., 2014b). This study found resveratrol not only disrupted Wnt signaling, but also abrogated STAT3 signaling.
Fig.1.Mechanisms of inhibitors within the Wnt pathway.Wnt inhibitors act at various points within the active Wnt pathway.Common targets include Wnt ligands, including sequestration by OMP-54F28, and the β-catenin/TCF interaction. LGK974 is unique in that it inhibits pathway activation by preventingWnt ligand secretion by inhibiting palmitoylation by PORC. COX inhibition by NSAIDS prevents PGE2 from blocking the function of GSK-3 and Axin. Other targets are theWnt receptor, Fzd, and co-receptor LRP5/6. Several inhibitors act to stabilize the destruction complex, thus preventing the accumulation of β-catenin and transcription of downstream effectors. Alternatively, others prevent transcription by inhibiting transcriptional co-factors.
Small molecule inhibitors
There are many inhibitors that speciﬁcally disrupt the interaction of β-catenin with other key components of the Wnt pathway. PNU74654 was discovered by high-throughput screening, and was shown to inhibit the interaction of β-catenin and TCF (Fig. 1 [Trosset et al., 2006]). Treatment with certain 2,4-diamino-quinazoline derivatives, another compound that disrupts the β-catenin/TCF interaction, resulted in 20– 35% tumor growth inhibition when colorectal cells were implanted in nude mice(Chen et al., 2009c). Another approach is to disrupt a different β-catenin/activator interaction. Emami et al. identiﬁed a small molecule inhibitor using a cellbased screen that speciﬁcally bound to CREB binding protein (CBP), a TCF co-activator, termed ICG-001 (Fig. 1 [Emami et al., 2004]). This inhibitor has been experimentally explored in other diseases with aberrant Wnt signaling including kidney disease and pulmonary ﬁbrosis with promising success (Hao et al., 2011; Henderson et al., 2010; Sasakiet al., 2013). ICG-001, at higher doses,was found to induce apoptosis in colon cancer cells with minimal effects on normal colon cells in vitro (Emami et al., 2004). Using ICG-001 in combination with a Met inhibitor and a CXCR4 inhibitor delayed tumor onset in a breast cancer mouse model (Holland et al., 2013). Tumors that arose from CSCs (CD24+ CD29+) isolated from salivary gland tumors grown in NOD/SCID mice and passed into NOD/SCID mice had decreased tumor volume when treatedwith ICG-001(Wend et al., 2013).These salivary gland tumors were originally grown in mice that were double-mutants, wherein mice had a gain of function mutation in β-catenin and a loss of function mutation in BMPR1A, a receptor in bone morphogenetic proteins (BMP) signaling which has been shown to inhibit CSCs proliferation in glioblastomas (Piccirillo et al., 2006). These were subsequently implanted into NOD/SCID mice, suggesting that Wnt inhibition with ICG-001 is effective in inhibiting tumor growth where Wnt activation is one of the key drivers. ICG-001 has also been shown to inhibit cell proliferation in pancreatic ductal adenocarcinoma(PDAC) by causing a G1 cell cycle arrest; however this effect appeared to be independent of Wnt signaling inhibition, suggesting ICG-001 may also target other pathways (Arensman et al., 2014). Several small molecule inhibitors, including XAV939, JW55 and IWR-1 promote β-catenin degradation by inhibiting PARsylation by Tankyrase 1 and Tankyrase 2 and thereby stabilizing axin (Fig. 1 [Chen etal.,2009a;Huangetal.,2009;Waaleretal.,2012]). XAV939 promoted Axin apoptosis in neuroblastoma cells and inhibited proliferation under serum-deprivation in breast and colorectal cancer cells (Bao et al., 2012; Tian et al., 2013). JW55 inhibited in vivo tumor growth in APC mutant mice using colorectal carcinoma cells (Waaler et al., 2012). Similarly, IWR-1 inhibited colon and prostate cancer cell growth (Chen et al., 2009a). Other small molecule inhibitors target Dvl or PORC (Fig. 1).LGK974 inhibits PORC, an O-acyltransferase that is required for the palmitoylation of Wnt ligands and ligand secretion (Zhai et al., 2004) and induced tumor regression in vivo using a mouse model for Wntdriven breast cancer and HNSCC (Liu et al., 2013). Interestingly exome sequencing a panel of 40 HNSCC cell lines showed a strong correlation between LGK974 sensitivity and Notch1 mutations although the significance of this has yet to be elucidated. Several small molecule inhibitors target the Wnt pathway by interacting with Dvl and thereby the destruction complex, ultimately leading to decreased β-catenin. NSC668036, FJ9 and 3289–8625 were shown to inhibit Wnt signaling by directly binding the PDZ domain of Dvl (Fujii et al., 2007; Grandy et al., 2009; Shan et al., 2005). 3289–8625 was shown to inhibit PC3 prostate cancer cell growth (Grandy et al., 2009), and FJ9 was shown to induce apoptosis in both melanoma and NSCLC cell lines (Fujii et al., 2007). FJ9 also inhibited tumor growth using implanted NSCLC cells in a mouse xenograft model. Two FDA-approved anthelmintics effectively inhibit the pathway by targeting several factors. Using a high-throughput small molecule screen, Pyrivinium was identiﬁed as a Wnt antagonist (Thorne et al., 2010). Pyrivinium, classically used in the treatment for pinworm infection (Royer & Berdnikoff, 1962), inhibits Wnt signaling at multiple points in the pathway (Fig. 1). It binds and induces a conformational change in CK1, promoting its kinase activity, and thus stabilizing axin and retaining β-catenin in the cytoplasm. Furthermore, it promoted the degradation of pygopus, a nuclear factor that is required by β-catenin for transcription of downstream Wnt targets (Thorne et al., 2010). Pyrivinium has recently been shown to target Wnt signaling in colon cancer cells, resulting in increased cell death, inhibition of cell migration and delaying liver metastasis growth in vivo (Wiegering et al., 2014). Another FDA-approved drug termed niclosamide, routinely used in the treatment of tapeworm, inhibited Wnt signaling by causing Fzd1 receptor internalization and decreased Dvl2 protein levels in human osteosarcoma cells (Fig. 1 [Chen et al., 2009b]). In contrast, another study suggests that niclosamide acts through targeting LRP6, both decreasing its phosphorylation and overall protein expression. In this study, Dvl2 was unperturbed, and decreased cell proliferation and apoptosis induction were observed in prostate and breast cancer cells (Lu et al., 2011b). These ﬁndings suggest the mechanisms are dependent on cell type, warranting more studies on this compound. Niclosamide was found to decrease spheroid growth, increase apoptosis and inhibit tumor growth in NOD/SCID mice when using the side population sorted from breast cancer cells (Wang et al., 2013). Both spheroid growth and side population highly enrich for CSCs, indicating that niclosamide may function in target CSCs. Ye et al. used breast cancer cells and observed decreased proliferation, migration and invasion, as well as increased apoptosis and decreased tumor growth in an in vivo mouse model (Ye et al., 2014). Niclosamide has also been used to target basal-like breast,liver,brain and ovarian cancer (Arend et al., 2014; Londono-Joshi et al., 2014; Tomizawa et al., 2013; Wieland et al., 2013; Yo et al., 2012). It is important to note that these in vivo studies, as well as the ones stated earlier, have shown little to limited levels of toxicity, providing hopeful optimism for Wnt inhibition in human cancer therapy.
Numerous studies have used viral-based targeting with recombinant adenoviruses (Barker & Clevers, 2006). This is accomplished by integrating TCF binding sites into robust promoters and thus achieving
cell killing speciﬁc to cells with active Wnt signaling. Cancer cell killing was attained through manipulation of adenoviruses with E1 and E2 promoters, and these effectively targeted cancers with aberrant Wnt signaling (Brunori et al.,2001; Fuerer & Iggo, 2002). Surprisingly, Brunori et al. observed cell killing in lung cancer cells, and little effect in colon cancer cells. However, the reason for this was largely unknown. Variations of this have been done, and in one study, the addition of ADP cytosolic protein boosted the ability of the virus to spread from cell to cell(Toth et al., 2004). In addition, viral expression and effectiveness in tumor growth inhibition using mouse xenograft models were speciﬁc to colon cancer cells and non-effective in lung cancer cells. This suggests speciﬁcity to those cancers with greater levels of Wnt activity. In another approach, using TCF-driven E1 and E4 promoters, the Na/I symporter (hNIS) gene was included in the recombinant adenovirus (Peerlinck et al., 2009). This resulted in the enhancement of 131I− radiotherapy and allowed for imaging and tracking the spread of the adenovirus using computed tomography(CT)imagingwhen injected with 99mTcO4 −. Similarly, other studies combine cytotoxic gene expression with speciﬁcity to the Wnt pathway by the integration of promoters that are under the control of TCF. Using this technique and various promoters, apoptosis promoting Fas-associated via death domain (fadd), diphtheria toxin A (DTA) and herpes simplex virus thymidine kinase (HSV TK) genes have all been expressed and shown to be effective in targeting cancer cells with active Wnt signaling (Chen & McCormick, 2001; Kwong et al., 2002; Lipinski et al., 2004). Alternatively, others have added a gene that enhances cytotoxicity of prodrugs in order to increase therapeutic efﬁcacy (Fuerer & Iggo, 2004; Lukashev et al.,2005).
As stated earlier, because of the overexpression of Wnt ligands and/or receptors in many cancers, antibody-based inhibitors have been developed to bind and sequester either free ligand or Fzd receptors (Fig. 1). Several antibodies toward Wnt ligands have been produced. A monoclonal Wnt-1 antibody was shown to induce apoptosis in NSCLC and breast cancer cells by Wnt inhibition and activating cytochrome c and caspase 3, as well as decreasing survivin expression (He et al., 2004). Furthermore, the Wnt-1 antibody inhibited tumor growth in nude mice with NSCLC cells implanted subcutaneously, independent of whether the antibody was administered at implantation or once tumors were established, suggesting the timing of antibody administration was irrelevant for tumor control. Similar results were observed in colon cancer and sarcoma using theWnt-1antibody(Heetal.,2005;Mikamietal.,2005). The Wnt-1 antibody also induced apoptosis in mesothelioma cells that weredeﬁcientin β-catenin, suggestingnon-canonical Wnt signaling inhibition was possible as well (You et al., 2004a). By the same token, a monoclonalWnt-2 antibody induced apoptosis in NSCLC and melanoma, as well as inhibited tumor growth in a melanoma xenograft model (You et al., 2004b, 2004c). In Wnt-activated HNSCC cells, both Wnt-1 and Wnt-10b antibodies effectively blocked Wnt signaling, induced apoptosis and inhibited cell proliferation (Rhee et al., 2002). OMP-18R5 is a monoclonal antibodythat was initially identiﬁed for its ability to bind Fzd7. Since then, OMP-18R5 has been found to bind Fzd1, Fzd2, Fzd5, Fzd7 and Fzd8 and block β-catenin signaling in responseto Wnt3a ligand (Gurney et al., 2012). In the same study, using humantumorxenografts,OMP-18R5inhibitedtumorgrowthinseveral tumor types, including colon, breast, pancreatic and lung cancers. Tumor recurrence was also delayed. Furthermore, the addition of OMP-18R5 to standard-of-care chemotherapies, such as paclitaxel, increased efﬁcacy in tumor growth inhibition in a synergistic manner. Although off-target effects were suggested in that several Wnt genes were inhibited in the mouse liver, at effective doses there was little toxicity observedin the GI tract. In another approach, peptides with complementary sequences interacting with either the Wnt ligand or Fzd receptor are fused with the immunoglobulin Fc domain. Using this approach, for example, WIF1-Fc and SFRP-Fc were expressed in cancer cells using recombinant adenoviruses (Hu et al.,2009).Wnt signaling inhibition by these antagonists inhibited tumor growth and prolonged survival in hepatocellular xenografts.
Effective Wnt targeting has been accomplished using an immunoglobin Fc fused to Fzd8, Fzd8(1–173)hFc (Fig. 1 [Hsieh et al., 1999b; Reya et al., 2003]). Others improved upon this, and constructed a minimal Fzd8 protein (residues1–155), wherein possible protease cleavage sites were removed (DeAlmeida et al., 2007). The fusion of the CRD domain of Fzd8 with Fc (F8CRDhFc) exhibited an extended half-life in vivo in comparison to Fzd8(1–173)hFc and successfully inhibited growth in human teratoma tumor xenografts with very limited toxicity to regenerating tissues. OMP-54F28 is a truncated Fzd8 receptor fused to the IgG1Fc region. This inhibitor has been shown to block Wnt signaling and block tumor growth using a MMTV-Wnt1 induced tumor model (Hoey, 2013). Furthermore,OMP-54F28 was shown to synergize with chemotherapeutic agents.When a patient-derived pancreatic cancer xenograft model was treated with gemcitabine and OMP-54F28, OMP-54F28 alone reduced tumor growth to a greater extent than gemcitabine alone and a combination of the two gave a slight advantage over single-agen tOMP-54F28. OMP-54F28 also reduced the frequency of CSCs as quantitated by the number of tumors that regrew when serially passaged (30, 90 or 270 cells) into NOD/SCID for 82 days. Similar to tumor growth inhibition, the greatest reduction in CSC frequency occurred in combination, and this was slightly greater than OMP-54F28 alone. However, with gemcitabine alone the frequency of CSCs increased when compared to control. The percent of cells expressing CD44+, a marker for CSCs, decreased from ~12.7% to 1.9%with OMP-54F28 treatment alone as compared to an increase from ~12.7% to 13.9% when treated with gemcitabine alone and from ~12.7% to 1.7% when a combination of OMP-54F28 and gemcitabine was used. Using luciferase-labeled pancreatic tumor cells implanted orthotopically, tumors were grown for 30 days, treated with OMP-54F28 and imaged in vivo for metastases. A decrease in both liver and lung metastases was observed (Hoey, 2013). Althoughcancers,including pancreatic cancers, are initially sensitive to gemcitabine, they can become resistant to treatment. A gemcitabine resistant pancreatic tumor model was created by continuously passing cells inincreasing concentration of gemcitabine. Using this gemcitabine resistant model, tumor growth was inhibited with a combination of 5FU and irinotecanor OMP-54F28 alone in comparison to control. However when all three are combined, there is a greater effect in tumor growth inhibition. Epithelial speciﬁc antigen (ESA)+CD201+, a marker of pancreatic CSCs, was assessed and a substantial decrease was observed with a combination of the three compounds, while treatment with OMP-54F28 alone showed the greatest decrease in ESA + CD201+. Treatment of gemcitabine resistant xenografts with OMP-54F28, gemcitabine and Abraxane resulted in tumor growth inhibition, and this growth inhibition was greater than that with the combination of gemcitabine and Abraxane (Hoey, 2013). Together the data suggest that OMP54F28 inhibits tumor growth, limits CSC frequency and tumor recurrence and is active in gemcitabine resistant tumors. It is effective as a single agent, but also in combination with chemotherapeutic agents.
OMP-54F28: ﬁrst-in-human clinical data
With the efﬁcacy seen in preclinical solid tumor models, OMP-54F28 has been recently investigated in a ﬁrst-in-human phase1a study with advanced solid tumors (Jimeno, 2014). The primary objective of this study was to determine the safety and toxicity proﬁle of the drug in patients with advanced solid tumors. Secondary objectives included pharmacokinetics, immunogenicity, and preliminary efﬁcacy of OMP-54F28. The study was designed as a 3+3 dose escalation trial with dose levels between 0.5 and 20 mg/kg given intravenously every 3 weeks. Dose limiting toxicities (DLTs) were assessed every 28 days, and tumor assessment was done every eight weeks. At the time of submission of this review, only preliminary data from the phase1 study has been reported. The main adverse effects seen with OMP-54F28 included dysgeusia, fatigue, muscle spasms, decreased appetite, nausea and vomiting. Modulation of the WNT pathway has been shown to have effects in the bone including bone remodeling (Goldring & Goldring, 2007). In the present study, β-C-terminal telopeptide (β-CTX), a marker of increased bone turnover, was closely assessed, and it was recommended that zoledronic acid should be given to patients with doubling of their β-CTX levels.
Ongoing studies with OMP-54F28
There are 3 ongoing phase 1b studies combining OMP-54F28 with other drugs in solid tumors based on preclinical data and the safety and tolerability found in the phase 1a trial (Table 2 [OncoMed Pharmaceuticals Inc., 2014]). The ﬁrst trial is combining OMP-54F28 with sorafenibin patients with hepatocellular cancer. Patients included must have locally advanced or metastatic hepatocellular cancer with no prior systemic therapies. Patients will receive sorafenib 400 mg orally twice daily with OMP-54F28 intravenously on day 1 of a 21-day cycle. Initiallydosesof5 mg/kgor10 mg/kgwillbeused,andbasedonsafety data higher or lower doses may be evaluated. The primary objectives are to evaluate the safety and tolerability of OMP-54F28 in combination with sorafenib, to identify dose limiting toxicities (DLTs) and maximum tolerated dose (MTD) and to determine the recommended phase 2 dose. Secondary objectives include characterization of the pharmacokinetics of OMP-54F28 in combination with sorafenib, characterization of the immunogenicity of OMP-54F28 and preliminary assessment of efﬁcacyof the two drugs combined. Another phase1b study will be evaluating the combination of OMP54F28, nab-paclitaxel,and gemcitabine in patients with pancreatic cancer. Patients enrolled must have previously untreated stage IV ductal adenocarcinoma of the pancreas. They will receive nab-paclitaxel 125 mg/m2 and gemcitabine 1000 mg/m2 intravenously on days 1, 8 and 15 of a 28-daycycle. OMP-54F28 will be given at either 3.5 mg/kg or 7.0 mg/kg intravenously on days 1 and 15. Depending on emerging safety data, higher doses may also be evaluated. The primary objectives of this study will include evaluation of the safety and tolerability of the drug combinations, identiﬁcation of the DLTs and MTD and identiﬁcation of the recommended phase 2 dose for OMP-54F28 in combination with nab-paclitaxel and gemcitabine. Secondary objectives will be characterization of the pharmacokinetics and immunogenicity of the drug combinations and preliminary assessment of the efﬁcacy of these drugs for metastatic pancreatic cancer. The third phase1b trial open is studying OMP-54F28 in combination with paclitaxel and carboplatin in ovarian cancer. The patients included should have recurrent, platinum-sensitive ovarian cancer, deﬁned as disease progression greater than 6 months after completing a minimum of 4 cycles of a platinum-containing chemotherapy regimen. Patients who have received prior treatment with paclitaxel and carboplatin for recurrent disease will be excluded. Paclitaxel 175 mg/m2 and carboplatin AUC 5 will be given intravenously on day 1 of a 21-day cycle. OMP-54F28 will be given at 5 mg/kg or 10 mg/kg intravenously on day 1 with potential further dose escalation based on safety data. The paclitaxel and carboplatin will be given for a maximum total of 6 cycles with OMP-54F28 continuing until disease progression.The primary objectives are safety and tolerability of OMP-54F28 in combination with paclitaxel and carboplatin, determination of any DLTs and the MTD and planned phase2 dose of OMP-54F28. Secondary objectives will include pharmacokinetics, pharmacodynamics and efﬁcacy of the drug combination.
7.11.2 Wnt signaling and hepatocarcinogenesis – Molecular targets
Hepatocellular carcinoma (HCC) is one of the most common causes of cancer death worldwide. HCC can be cured by radical therapies if early diagnosis is done while the tumor has remained of small size. Unfortunately, diagnosis is commonly late when the tumor has grown and spread. Thus, palliative approaches are usually applied such as transarterial intrahepatic chemoembolization and sorafenib, an anti-angiogenic agent and MAP kinase inhibitor. This latter is the only targeted therapy that has shown significant, although moderate, efficiency in some individuals with advanced HCC. This highlights the need to develop other targeted therapies, and to this goal, to identify more and more pathways as potential targets. The Wnt pathway is a key component of a physiological process involved in embryonic development and tissue homeostasis. Activation of this pathway occurs when a Wnt ligand binds to a Frizzled (FZD) receptor at the cell membrane. Two different Wnt signaling cascades have been identified, called non-canonical and canonical pathways, the latter involving the β-catenin protein. Deregulation of the Wnt pathway is an early event in hepatocarcinogenesis and has been associated with an aggressive HCC phenotype, since it is implicated both in cell survival, proliferation, migration and invasion. Thus, component proteins identified in this pathway are potential candidates of pharmacological intervention. This review focuses on the characteristics and functions of the molecular targets of the Wnt signaling cascade and how they may be manipulated to achieve anti-tumor effects.
HCC represents a major public health problem with a high impact on society. HCC is the sixth most common tumor worldwide in terms of incidence (about one million per year). Projections are that this incidence will substantially increase during the next decades due to persistent infection with the hepatitis C virus as well as the emergence of non-alcoholic steatohepatitis as a major health problem. HCC portends a poor prognosis since ranking third in terms of “cause of death” by cancer, and often presents as a major complication of cirrhosis related to chronic hepatitis B and C infections, or non-virus related [, , ]. The dismal prognosis is generally related to a late diagnosis after HCC cells have infiltrated the liver parenchyma, have spread through the portal venous system and/or have formed distant metastases. However, if HCC is diagnosed early (<20% of patients), these smaller tumors may be cured by surgical resection, liver transplantation or radiofrequency ablation. In more advanced tumors (>80% of patients at diagnosis), only palliative approaches can be applied. In this regard, transarterial intrahepatic chemoembolization has been shown to be somewhat effective in increasing overall survival of individuals with tumors that have spread only into the liver parenchyma without extrahepatic metastasis (median overall survival is increased from 15 to 20 months compared to the best supportive care). In HCC with extrahepatic spread, only sorafenib, an anti-angiogenic and MAP kinase inhibitor, has been shown to increase overall survival of patients (from 8 to 11 months) . All other systemic approaches such as cytotoxic chemotherapy have not been shown to be effective; thus, to date, no targeted therapy except sorafenib has been proven to prolong life in patients with HCC. However, there are ongoing or ended clinical trials with agents that target FGF, VEGF, PDGF, EGF, IGF, mTOR, and TGFβ signaling pathways but none has been shown yet to have a significant impact on patient survival .
Recently, cancer stem cells (CSC) have been hypothesized to play a key role in tumor maintenance as well as relapse after surgical resection. There is accumulating information that supports a role for CSC in hepatocarcinogenesis to maintain the tumor size and to initiate tumor recurrence following therapy . The pool of CSC is maintained by self-renewal capabilities that are largely driven by reactivation of embryonic signaling programs mediated by Wnt, Notch, Bmi, and Hedgehog pathways, similar to what has been previously demonstrated during breast carcinogenesis . Preclinical studies further underline the potential value of inhibiting activation of these signaling programs in some tumor types [,, , ].
In this review, we describe the features of a therapeutic target, i.e., the Wnt pathway, for potential therapy of HCC. We will discuss experimental and preclinical studies regarding the use of Wnt inhibitors as a therapeutic approach for HCC.
The Wnt-mediated signaling
The first member of the Wnt family of ligands was identified from the int-1 gene found in a mammary adenocarcinoma, located at the integration site of the mouse mammary tumor virus (MMTV); subsequently, it was demonstrated to have oncogenic properties . More important, int-1 homolog genes have been found in human tumors as well . In addition, a highly conserved int-1 homolog was also discovered in Drosophila and designated Wingless “Wg” . The combination of int-1 and Wingless led to the common Wnt1 terminology and recently has been used to designate the Wnt family of ligands .
Wnt proteins are secreted extracellular auto-paracrine glycoproteins that interact with Frizzled receptors (FZD), a seven transmembrane domain protein, resembling the G-protein-coupled receptor (GPCR) family. Vinson and colleagues revealed that FZD contains an extracellular cysteine-rich domain (CRD) which is the putative binding site for the Wnt ligands. These investigators demonstrated the functional role of the frizzled locus to coordinate development of the cytoskeleton in Drosophila epidermal cells . Subsequently, Wnt/FZD-mediated signaling has been extensively studied, and although it has been widely implicated in cellular homeostasis, these ligand/receptor interactions have now been appreciated as key factors during the oncogenesis process and therefore, could serve as new therapeutic targets.
Thus, Wnt proteins represent members of a highly conserved family that is involved in several processes including embryonic development, cell fate determination, proliferation, polarity, migration, and stem cell maintenance. In addition, Wnt/beta-catenin signaling has been found to play key roles in metabolic zonation of adult liver, regeneration . In adult organisms, deregulation of Wnt signaling may lead to tumor development [, ]. The Wnt-mediated pathway is activated through the binding of one Wnt ligand to a FZD receptor. Ten different FZD receptors and 19 Wnt ligands have been identified in humans. The binding of Wnt to an FZD receptor can trigger activation of at least three different pathways. The first is the Wnt/β-catenin cascade, also called the Wnt-canonical pathway; the remaining two are the planar cell polarity (PCP) and the Wnt/calcium pathways, respectively. The two latter are β-catenin independent and represent examples of the non-canonical cascades. In this regard, a multitude of combinations between the 19 Wnt ligands and the 10 FZD receptors, such as co-receptors and other molecules, are theoretically possible. Classically, Wnt1/2/3/3a/8a/8b/10a/10b and FZD1/5/7/9 are classified as the canonical elements, whereas Wnt4/5a/5b/6/7a/7b/11 and FZD2/3/4/6 are designated as non-canonical components. The remaining Wnt2b/9a/9b/16 and FZD8/10 proteins remain unclassified [, ]. However, it remains elusive how selectivity between Wnt/FZD as well as specificity of downstream signaling is achieved. Some Wnt/FZD elements can share dual canonical and non-canonical functions. For instance, it has been shown that in absence of Ror2 co-receptors, Wnt5a can activate β-catenin signaling with FZD4 and Lrp5 . FZD3 has been described to act likely through canonical pathways in mice neurogenesis . Zhang et al. demonstrated that in Xenopus foregut, FZD7 can activate low level of β-catenin and non-canonical JNK signaling in which both pathways contributed to foregut fate and proliferation while JNK pathway regulated cell morphology . It is of interest that canonical and non-canonical pathways can not only be driven by specific Wnt/FZD combinations, but also by cell type, differentiation status, localization and composition of the microenvironment .
The canonical Wnt/FZD pathway
The β-catenin protein, encoded by the CTNNB1 gene, is a key component of Wnt-canonical pathway signaling. β-catenin has a central region which presents armadillo domain repeats important for the binding of partners, such as Axin1 and adenomatous polyposis coli protein (APC) as well as transcription factors . The C- and N-terminal regions are important. C-terminus of β-catenin serves as a binding factor for a multitude of complexes promoting β-catenin-mediated transcription, whereas phosphorylation of the N-terminus promotes degradation of β-catenin. Indeed, β-catenin may be present in several cellular compartments, such as the inner plasma membrane having a role in cell-cell junctions, the cytoplasm and the nucleus where it forms an active complex containing TCF/LEF transcription factors (T-cell factor/lymphoid enhancer factor) . In the absence of nuclear β-catenin, TCF/LEF interact with the transcriptional co-repressor transducin like enhancer-1 (TLE-1) (Drosophila homolog Groucho), thus preventing β-catenin target gene expression . Following translocation into the nucleus, β-catenin binds to TCF/LEF and replaces the TLE-1 repressor to form a transcriptional complex that activates the expression of its target genes (Fig. 1).
Canonical Wnt/FZD signaling pathway
Fig. 1 Canonical Wnt/FZD signaling pathway. (A) In the absence of Wnt signaling, soluble β-catenin is phosphorylated by a degradation complex consisting of the kinases GSK3β and CK1α and the scaffolding proteins APC and Axin1. Phosphorylated β-catenin is targeted for proteasomal degradation after ubiquitination by the SCF protein complex. In the nucleus and in the absence of β-catenin, TCF/LEF transcription factor activity is repressed by TLE-1; (B) activation of the canonical Wnt/FZD signaling leads to phosphorylation of Dvl/Dsh, which in turn recruits Axin1 and GSK3β adjacent to the plasma membrane, thus preventing the formation of the degradation complex. As a result, β-catenin accumulates in the cytoplasm and translocates into the nucleus, where it promotes the expression of target genes via interaction with TCF/LEF transcription factors and other proteins such as CBP, Bcl9, and Pygo.
In absence of the canonical Wnt signaling, cytosolic β-catenin is targeted for degradation by a complex composed of a scaffold of proteins named axin1, APC, and two serine/threonine kinases: the glycogen synthase kinase 3β (GSK3β) and the casein kinase 1 (CK1)  (Fig. 1A). Axin1 and APC act together as scaffolding proteins through binding of β-catenin, and enhance its N-terminal phosphorylation by GSK3β and CK1. The first phosphorylation event is generated by CK1 at Ser45 which allows the GSK3β-mediated sequential phosphorylation of Thr41, Ser37, and Ser33 [, ]. Ser37 and Ser33 phosphorylations provide a binding site for the E3 ubiquitin ligase β-TRCP (β-transducin repeat containing protein), leading to β-catenin ubiquitination in a β-TRCP/Skp1/cullin F-box complex (SCF) dependent manner followed by proteasomal degradation [, ].
Activation of the canonical Wnt signaling cascade leads to disruption of the β-catenin degradation complex, resulting in β-catenin accumulation in the cytoplasm followed by translocation into the nucleus where it serves as a transcription factor to activate downstream target genes (Fig. 1B). In brief, this process is as follows: Wnt ligand binds to the extracellular domain of an FZD receptor and Lrp5/6 co-receptors. This ternary complex (Wnt/FZD/Lrp) recruits the scaffolding phosphoprotein dishevelled (Dvl/Dsh) at the plasma membrane which in turn traps the axin-bound-GSK3β complex, thus preventing proteasomal degradation of cytosolic β-catenin. When stabilized, β-catenin is able to translocate into the nucleus, where it binds to TCF/LEF transcription factors and then forms a transcriptionally active complex with pygopus (Pygo), CBP (CREB-Binding Protein) and Bcl9 proteins . In mammals, four TCF genes have been described, which adds further complexity to the mechanism(s) of activation of the Wnt canonical cascade . Of notice is the β-catenin pool localized at the plasma membrane that plays a key role in cell-cell junctions. To this aim, a complex including either p120 catenin/γ-catenin(plakoglobin)/α-catenin or p120 catenin/β-catenin/α-catenin  binds to the cytoplasmic carboxyl terminus domain of E-cadherin adhesion molecule, in order to join cadherins to the actin cytoskeleton. More precisely, p120 catenin binds to the juxtamembrane and then β-catenin or γ-catenin binds to the cytoplasmic domain of E-cadherin. The remaining α-catenin serves as a link between actin and β/γ-catenin which leads to the stabilization of cell adhesion . The possible consequences of inhibiting β-catenin at adherent junctions have to be discuss in respect of their role in epithelio-mesenchymal transition (EMT). Disruption of E-cadherin-mediated adherent junctions is a major event in EMT  and because of the interplay between cadherin-mediated cell adhesion and canonical/β-catenin signaling , targeting β-catenin could also promote the disruption of these junctions leading to enhance EMT. However, Wickline et al.have shown that in hepatocyte-specific β-catenin-conditional null mice, γ-catenin is upregulated and associated with E-cadherin and actin to maintain adherent junctions. In addition, no nuclear γ-catenin was detected in liver of KO mice, leading to the conclusion that despite armadillo domains on γ-catenin, there is no compensation at nuclear level. Nevertheless, authors warn us about preventing concurrent γ-catenin suppression that may increase tumor cell invasion . More recent study confirmed these results in in vitro experiments with HCC cell lines and identified the mechanism of γ-catenin stabilization as serine/threonine phosphorylation induced by protein kinase A . With regard to this recent data, targeting β-catenin in HCC therapies may not disturb cell junctions since the design of Wnt inhibitors for therapeutic intervention, specifically designates soluble active β-catenin as preferential target.
The non-canonical Wnt/FZD pathways
In contrast to the canonical Wnt pathway, non-canonical signaling does not depend on β-catenin and requires Ror2/Ryk co-receptors instead of Lrp5/6 (Fig. 2). In the Wnt/PCP pathway, Wnt/FZD interaction promotes the recruitment of Dvl/Dsh, which in turn binds to the small GTPase protein called Rac, leading to both the induction of ROCK (Rho-associated protein kinase) pathway and the activation of the MAP kinase cascade and subsequently to the activation of AP1-mediated target gene expression [, ]. In the Wnt/calcium pathway, the complex formation between FZD, Dvl/Dsh and G proteins results in PLC (Phospho Lipase C) activation which cleaves PIP2 (Phosphatidyl Inositol 4,5 biphosphate) into DAG (DiAcylGlycerol) and IP3 (Inositol 1,4,5-triphosphate). This process results in the activation of PKC (Protein Kinase C) through DAG while IP3 promotes calcium release from the endoplasmic reticulum. Increased intracellular concentration of calcium enhances phosphorylation and activation of PKCs. This also triggers the activation of Ca2+-calmodulin-dependent calcineurin and CAMKII (Ca2+-calmodulin dependent kinase II), leading to NFAT (Nuclear Factor of Activated T-cell) and NLK (Nemo Like Kinase) translocation, respectively. NLK acts as a β-catenin pathway inhibitor through phosphorylation and degradation of TCF/LEF transcription factors .
Non-canonical Wnt/FZD signaling pathway
Fig. 2 Non-canonical Wnt/FZD signaling pathways. Interaction of Wnt, FZD, and ROR2/RYK co-receptors leads to either (1) JNK activation, (2) PKCs activation, (3) NFAT transactivation, or (4) inhibition of β-catenin activity through binding of NLK to TCF/LEF.
Antagonists and agonists of Wnt/FZD-mediated signaling
Several secreted proteins are known to negatively or positively regulate the Wnt/FZD complex. Four classes of antagonistic molecules have been described. Wnt inhibitory protein-1 (Wif1) and secreted FZD-related proteins (sFRP1, 2, 3, 4, 5) bind to and sequester the soluble Wnt ligands, thus inhibiting their interaction and binding to FZD receptors [,, , ]. The Dickkopf family is composed of four members (Dkk1, 2, 3, 4) that can interact with both Lrp5/6 and Krm1,2 (Kremen1,2) co-receptors . The ternary complex Lrp-Dkk-Krm prevents β-catenin stabilization by promoting Lrp5/6 endocytosis . Wise and Sost proteins form the other class of secreted antagonists. They bind to Lrp5/6 and thus disrupt the Wnt-induced FZD-Lrp5/6 interaction [, ].
Three agonistic molecules have recently been identified; the R-spondins (Rspo1, 2, 3, 4), norrin and glypican-3 (Gpc3). The Gpc3 is a heparan sulfate proteoglycan bound to the cell membrane through a glycosyl-phosphatidylinositol anchor. Gpc3 increases autocrine/paracrine canonical Wnt signaling by binding to Wnt ligands, thus facilitating the interaction between Wnt ligands and FZD receptors . Mechanisms by which Rspo and Norrin activate the canonical Wnt pathway have not been clarified. Rspo1 is able to bind to both Lrps and FZDs but it has also been proposed that Rspo prevents Lrp6 internalization through binding to Krm instead of Dkk [, , ].
Wnt signaling deregulation in human hepatocarcinogenesis
Similar to other tumor tissue types, the canonical Wnt/FZD signaling is a critical contributor to HCC pathogenesis. Indeed, 40–70% of HCCs harbor nuclear accumulation of the β-catenin protein, one of the hallmarks of the Wnt/β-catenin pathway activation [, , ]. Activating mutations of the β-catenin gene (CTNNB1) occur in 8–30% of tumors, while loss-of-function/mutations in APC and Axin genes occur in 1–3% and 8–15%, respectively and are mutually exclusive to CTNNB1mutations [, , , , , , ]. Some observations suggest that the CTNNB1 mutation could be a late event during hepatocarcinogenesis. However, accumulation of β-catenin was detected in the early stage of HCC development, suggesting that other mechanisms could contribute to β-catenin stabilization (Table 1) [, ]. Strikingly, extrinsic activation of Wnt/β-catenin pathway and CTNNB1 mutation do not lead to the same molecular expression pattern, supporting different roles for wild type and mutated β-catenin. The Wnt/β-catenin activated HCC subclass with a CTNNB1mutation is characterized by upregulation of liver-specific Wnt-targets, low grade and well-differentiated tumors, with chromosome stability and a favorable prognosis. The Wnt/β-catenin activated HCC subclass without CTNNB1 mutation is characterized by dysregulation of classical Wnt targets, high chromosomal instability, aggressive phenotype, and is preferentially associated with chronic HBV infection [, , ].
|Table 1Most prevalent potential mechanisms involved in activation of beta-catenin found so far in HCCs.|
Modulation of Wnt ligands or FZD receptor expression could account for Wnt/β-catenin pathway activation without any other mutations in CTNNB1, APC, or Axin genes. Indeed, upregulation of activators, such as ligands (Wnt1/3/4/5a/10b) or receptors/co-receptors (FZD3/6/7, Lrp6), and downregulation of inhibitors (sFRP1/4/5, Wif1, Dkk3, Dkk4) have been reported both in HCC tumors and surrounding precancerous liver tissues, which emphasizes that their over and/or underexpression may be early molecular events during hepatocarcinogenesis [, , , , , , ,
Modulation of Wnt ligands or FZD receptor expression could account for Wnt/β-catenin pathway activation without any other mutations in CTNNB1, APC, or Axin genes. Indeed, upregulation of activators, such as ligands (Wnt1/3/4/5a/10b) or receptors/co-receptors (FZD3/6/7, Lrp6), and downregulation of inhibitors (sFRP1/4/5, Wif1, Dkk3, Dkk4) have been reported both in HCC tumors and surrounding precancerous liver tissues, which emphasizes that their over and/or underexpression may be early molecular events during hepatocarcinogenesis [, , , , , , , ,].
Although β-catenin activation is crucial for liver development and regeneration, it is not sufficient per se for initiation of hepatocarcinogenesis. Indeed, animal models overexpressing an active β-catenin protein do not spontaneously form HCC [, , ]. However, β-catenin activation may cooperate with other oncogenic pathways such as insulin/IGF-1/IRS-1/MAPK, H-RAS, MET, AKT and chemicals to induce HCC formation in mice [, , , ]. It is described that beta-catenin mutation is a late event in hepatocarcinogenesis since present in some HCC tumors whereas absent in preneoplastic lesions, thus prompting us to speculate that only non-mutated beta-catenin could play a role in very early steps of hepatocarcinogenesis such as initiation and promotion. However, mutated forms of beta-catenin are used in experimental models to assess the role of activated beta-catenin in hepatocarcinogenesis. In these experimental mouse models, it is well shown that mutated beta-catenin is insufficient alone and per se for initiation of HCC but only enhance tumor promotion either in a context of chromosomal instability and increase of susceptibility to DEN-induced HCC formation [, ], or in a context of Lkb1+/− mice that spontaneously develop multiple hepatic nodular foci (NdFc) followed by HCC , or in a context of H-Ras transgenic mice where mutated beta-catenin appears as a strong carcinogenic co-factor collaborating with the mutated Ras oncogene . In contrast and apparently paradoxically, invalidation of beta-catenin in hepatic beta-catenin conditional knockout mice has been found as enhancing DEN-induced tumorigenesis . Of interest is another model of HCC developing in mice under exposure to phenobarbital (PB, potent tumor promoter in mouse liver) and DEN as tumor initiator. A tumor initiation–promotion study was conducted in mice with conditional hepatocyte-specific knockout (KO) of Ctnnb1 and in Ctnnb1 wild type controls. As expected, DEN + PB strongly enhanced liver tumor formation in Ctnnb1 wild type mice. Amazingly, the prevalence of tumors in Ctnnb1 KO mice was 7-fold higher than in wild type mice, suggesting an enhancing effect of the gene KO on liver tumor development . Thus there is a paradox where the absence of wild type beta-catenin or presence of the mutated form, both lead to enhanced DEN-induced hepatocarcinogenesis. The issue is that the discussion is speculative since the mechanism of increased HCC in conditional beta-catenin KO is unknown. In the design of Wnt inhibitors for therapeutic intervention, these agents do target the Wnt pathway through the soluble beta-catenin cascade, but do not impact on invalidation of the beta-catenin pool involved in the membrane catenin/cadherin complexes involved in cell homeostasis. The beta-catenin therapeutic targeting may need to be personalized, based on the unexpected findings of enhanced tumorigenesis after chemical exposure in hepatocyte-specific beta-catenin conditional knockout mice.
Although the role of Wnt/β-catenin pathway is debated with respect to the initiation of hepatocarcinogenesis, it is definitively implicated in determining HCC aggressiveness, due to its promotion of increased cell proliferation, migration and invasion. This finding has been further substantiated by ectopic expression of Wnt3 and FZD7, Lrp6 or downregulation of sFRP1, Dkk1 and Dkk4 in HCC cell lines [, , , , ]. Moreover, recent studies have revealed that the Wnt/β-catenin pathway is also involved in the self-renewal and expansion of HCC initiating cells (i.e., the so-called liver CSC) which also influences tumor aggressiveness and resistance to chemo- radio-therapeutic agents [,]. Furthermore, Wnt/FZD-mediated signaling could influence tumor microenvironment that supports tumor survival, growth, and size. Recent investigations emphasize the role of sFRP1 in the induction of senescence of tumor-associated fibroblasts after chemotherapeutic treatment [, , ].
It is noteworthy that the canonical and non-canonical Wnt/FZD pathways may have complementary roles in the pathogenesis of HCC. Indeed, β-catenin activation appears to be involved in the tumor initiation phase of hepatic oncogenesis, whereas subsequent activation of non-canonical pathways associated with inactivation of β-catenin may enhance tumor promotion and progression . However, non-canonical pathways can also exhibit opposite effects on tumor behavior, since specific Wnt/FZD combinations are able to function as tumor suppressors . Although little is known about the role of Wnt/PKC pathway in HCC, it has been demonstrated that inhibition of PKCβ activity reduces motility and invasion properties of HCC cells . Finally, activation of the Wnt/JNK pathway during HCC progression would presumably support tumor growth, since enhanced JNK activity appears to be involved in HCC cell proliferation both in vitro and in vivo .
Identification of molecular targets for therapeutic interventions
There is some evidence to link the Wnt pathway activation to tumor cell properties characteristic of the malignant phenotype, such as enhanced cell proliferation, migration and invasion, which raises the possibility to target members of this signaling cascade as an attractive therapeutic approach for treatment of HCC [, ] (Fig. 3).
Potential Wnt-component targets
Fig. 3 Potential Wnt-component targets for therapeutic intervention on tumor development and growth. Inactivation of Wnt signaling pathway could be achieved by: (1) targeting extracellular signaling molecules with monoclonal antibodies, soluble factors or small molecules; (2) preventing the FZD/Dvl interaction; (3) stabilizing the destruction complex or (4) increasing β-catenin proteasomal degradation and (5) preventing the interaction between β-catenin and its co-factors for transactivity in the nucleus. The relationship between therapeutic molecules and their protein targets is indicated by a color code. Molecules in bold have been tested in HCC model, those in italics in other models of tumor growth.
Targeting extracellular molecules of the Wnt pathway
Antibody-based therapies directed against the overexpressed Wnt ligands and FZD proteins could provide a therapeutic approach. For instance, preclinical experiments have shown that an anti-Wnt1 monoclonal antibody inhibits the Wnt signaling pathway resulting in enhanced apoptosis and inhibiting cell proliferation, both in vitro and in vivo in a xenograft model of HCC . These findings have been experimentally validated for several other types of tumors, such as sarcomas, colon, breast, non-small-cell lung cancer, and head-neck squamous cell carcinomas [, , , ]. Interestingly, as demonstrated with a colon cancer cell line, this anti-Wnt antibody was able to induce apoptosis even in the presence of downstream mutations in APC or CTNNB1 genes and appeared to be synergistic with docetaxel chemotherapy with respect to therapeutic response . Although not tested in HCC tumors thus far, anti-Wnt2 antibodies may be useful to inhibit the Wnt/β-catenin cascade. Such antibodies induce apoptosis and inhibit tumor growth in vivo in several tumor types, including melanoma, mesothelioma, and non-small-cell lung cancer [, , ]. Since non-canonical pathways seem to be implied in tumor progression, the inhibition of Wnt-related ligand could be considered for therapy. For instance, WNT5A, which seems to be involved in the non-canonical pathway in HCC , could be antagonized by the use of anti-WNT5a antibodies. Indeed, in gastric cancer cells where WNT5A activates the non-canonical pathway, its inhibition reduces migration and invasion activities in vitro and in vivo . Nevertheless, since the non-canonical pathway could antagonize the canonical one, it might be deleterious to inhibit the former. Anti-FZD7 antibodies that induce apoptosis and decrease cell proliferation both in vitro and in vivo of FZD7 positive Wilms’ tumor cells are also available . More recently, a multispecific antibody that targets both FZD 1, 2, 5, 7, and 8 and mainly affects the canonical signaling pathway has been developed. It triggers a therapeutic reduction of breast, colon, lung and pancreas tumor growth and synergizes with other chemotherapeutic agents as well . Strikingly, this antibody remains effective even in tumor cells with APC or CTNNB1 gene mutations. In addition, FZD co-receptors could also be attractive targets for monoclonal antibody therapy since, in a retinal pigment epithelial cell line, anti-Lrp6 antibody has been shown to inhibit Wnt signaling .
Another therapeutic strategy would be to trap the endogenous Wnt ligands with the exogenous soluble form of FZD receptors. This approach was reported for FZD7 by Tanaka and colleagues in esophagus carcinoma cells and confirmed later in HCC cells [, ]. More recently, Wei and co-workers have developed the same approach using an FZD7 extracellular domain peptide (sFZD7) that can bind to and sequester the soluble Wnt3 ligand. This peptide decreased the viability of HCC cell lines with high specificity, since normal hepatocytes were not sensitive to sFZD7. Moreover, sFZD7 cooperates with doxorubicin to reduce HCC cell proliferation in vitro and in a xenograft murine model as well. Interestingly, it has been shown to be highly efficient and independent of the β-catenin mutational status . Inhibition of Wnt secretion by the small molecules, IWP2 and Wnt-C59 may also prevent autocrine Wnt signaling activation, as observed in colon cancer cell lines. These small molecules are also able to inhibit the progression of mammary tumors in Wnt1 transgenic mice [, ]. Addition of Wnt antagonist, such as sFRP1 or Wif1, has shown encouraging therapeutic results in HCC cell lines by blocking the Wnt/β-catenin signaling. These soluble molecules induce apoptosis, reduce angiogenesis and cell proliferation both in vitro and in vivo and are not influenced by the CTNNB1 mutation status . Other Wnt antagonists such as sFRP2 and sFRP5 should also be considered, since they show similar treatment effects in colon cancer as sFRP1 exhibits in HCC . Interestingly, Dkk1 and sFRP1 addition cooperates with anti-FZD7 antibodies to increase apoptosis in Wilm’s tumor demonstrating the importance of combinatorial therapies . Therapeutic small molecules, such as niclosamide and silibinin, display anti-tumor activity in vitro and in vivo by suppressing Lrp6 expression, leading to inhibition of Wnt/β-catenin signaling in human prostate and breast tumor cells, as well as by promoting induction of apoptosis [, ].
Targeting the Wnt-mediated pathway in the cytosol
The straight-in approach to inhibit Wnt/β-catenin pathway is to directly target β-catenin by small interfering-RNA or antisense based therapy, which can reduce cell proliferation and survival of HCC cell line, providing a proof of principle for this approach [, , ]. However, its potential use as a therapeutic tool remains unlikely since β-catenin protein is essential for cell junction. Thus, targeting the soluble active pool of β-catenin seems more appropriate.
The interaction between the cytosolic tail of FZD and its adaptor Dvl protein is of importance in mediating Wnt signaling. A proof-of-principle has clearly been established in HCC cells, by using small interfering peptides capable of entering the tumor cells and disrupting the interaction between a specific motif on the FZD7 cytosolic tail and the PDZ domain of Dvl . Similar results have been obtained in melanoma and non-small-cell lung cancer cells with small molecules using this same strategy .
Targeting the β-catenin destruction complex (APC, Axin, CK1, and GSK3β) as a therapeutic target has not been assessed in HCC so far. However, using other tumor model systems, such a strategy has demonstrated some potential. Since Axin1 overexpression induces apoptosis in HCC harboring APC, Axin1 or CTNNB1 mutations, stabilization of Axin1 would be an attractive approach to trigger β-catenin degradation . This may be achieved by using inhibitors of the Axin1 or/and 2 degradation, such as the smalls peptides IWR2, JW55 or XAV939 that inhibit the Wnt/β-catenin pathway, leading to a decreased proliferation of colon and breast cancer cell lines. Nevertheless, recent findings support the idea that this decrease may be restricted to low nutriment conditions, and emphasizes that stabilization of Axin needs to be combined with other therapeutic approaches [, , , ]. Preventing β-catenin stabilization through GSK3β activation would also be possible due to the discovery of differentiation-inducing factors (DIFs), which are natural metabolites expressed by Dictyostelium discoideum. Although the mechanisms of action of DIFs activity remain poorly understood, it is well known that DIFs induce β-catenin degradation and subsequently reduce cyclin D1 expression and function . CK1α, another component of the destruction complex, may be stabilized by pyrvinium that inhibits both Wnt signaling and cell proliferation, even in the presence of APC or CTNNB1 mutations, as observed in colon cancer cell lines . Another therapeutic approach would be to enhance β-catenin proteasomal degradation. In HCC, colon and prostate cancer cell lines, the small molecule antagonist CGK062 has been shown to exert such an effect, via the induction of β-catenin phosphorylation in the N-terminal domain which promotes its degradation . Two chemicals agents, hexachlorophene and isoreserpine, upregulate Siah-1, an ubiquitin ligase that induces β-catenin degradation, independent of its phosphorylation status, thereby inhibiting Wnt signaling and subsequently has been shown to reduce colon cancer cell proliferation .
Targeting the Wnt pathway in the nucleus
Finally, an alternative way to block Wnt-mediated signaling is to target the nuclear β-catenin per se and/or the co-factors responsible for transcription of downstream Wnt-responsive genes. To accomplish this aim, several small molecules have been identified. The FH535 agent prevents both Wnt- and PPAR- (Peroxisome Proliferator-Activated Receptors) mediated signaling by suppressing the recruitment of β-catenin co-activators to target gene promoters and has been shown to be active in HCC, colon, and lung tumor cell lines . PKF115-584, PKF118-310, and CGP049090 are inhibitors of TCF/β-catenin binding to DNA target sequences. They induce apoptosis in vitro and in vivo, as well as cell cycle arrest at the G1/S phase and suppress tumor growth in vivo independently of the mutated status of CTNNB1 . Furthermore, inhibition of β-catenin/CBP interaction by ICG-001 both selectively induces apoptosis in transformed, but not in normal colonic cells and reduces growth of colon carcinoma cells in vitro as well as in vivo . A second generation ICG-001 (PRI-724) is also available and in phase-I clinical trial (http://clinicaltrials.gov/show/NCT01302405). Other β-catenin binding proteins such as TBP, Bcl9, and Pygo also represent attractive approaches for inactivating Wnt signaling. Finally, interferon can inhibit β-catenin signaling through upregulation of RanBP3 that is a nuclear export factor, serving as extruding β-catenin outside the nucleus .
Conclusions and perspectives
Developmental regulated signaling pathways, such as Notch, Hedgehog and Wnt, have become important targets for new cancer drug development. While Notch and Hedgehog inhibitors are already in clinical trials, the Wnt inhibitors are still under preclinical assessment and only a few compounds have started to reach the phase-I clinical trials, since only recently has this pathway been recognized as playing a key role in tumor development. However, many studies have established proof-of-principle that specific targeting of molecules in this pathway can partially or fully switch off canonical as well as non-canonical Wnt signaling and lead to substantial anti-tumor activity. Thus, biotechnology and pharmaceutical organizations are currently developing Wnt signaling inhibitors. These inhibitors can target upstream or downstream proteins in this pathway. Targeting the Wnt cascade upstream of APC is controversial because downstream activating mutations in APC would, in theory, still drive tumor development. To cover the broadest number of activating mutations that occur in tumors, it seems that the ideal antagonist would be one that exerts its anti-tumor effect in the nucleus. Nevertheless, several experiments show that upstream targeting can also be very effective. Of importance is the potential toxicity of Wnt inhibitors on normal cells. Indeed, the Wnt pathway is critical for tissue and liver regeneration and for the ability of stem cells to self-renewal. Wnt pathway inhibitors could therefore have substantial and long-term side effects including anemia, immune suppression, as well as damage of the gastrointestinal tract. It is unknown what may occur in an adult mammal when this pathway is shut down or reduced in normal activity. Despite these known and unknown pitfalls, drug development is moving steadily forward to generate and characterize Wnt pathway inhibitors bothin vitro and in vivo. Indeed, agents that inhibit Wnt/β-catenin signaling as a means to produce anti-tumor effects are currently being assessed in clinical trials.
7.11.4 SALL4 is directly activated by TCF.LEF in the canonical Wnt signaling pathway
The SALL4 promoter has not yet been characterized. Animal studies showed that SALL4 is downstream of and interacts with TBX5 during limb and heart development, but a direct regulation of SALL4 by TBX5 has not been demonstrated. For other SAL genes, regulation within the Shh, Wnt, and Fgf pathways has been reported. Chicken csal1 expression can be activated by a combination of Fgf4 and Wnt3a or Wnt7a. Murine Sall1 enhances, but Xenopus Xsal2 represses, the canonical Wnt signaling. Here we describe the cloning and functional analysis of the SALL4 promoter. Within a minimal promoter region of 31 bp, we identified a consensus TCF/LEF-binding site.The SALL4 promoter was strongly activated not only by LEF1 but also by TCF4E. Mutation of the TCF/LEF-binding site resulted in decreased promoter activation. Our results demonstrate for the first time the direct regulation of a SALL gene by the canonical Wnt signaling pathway.
SALL4 is one out of ﬁve (four functional (SALL1-4) and one pseudogene (SALL1P)) human genes related to spalt (sal) of Drosophila melanogaster [1–5]. Mutations of SALL4 cause Okihiro/Duane-Radial Ray syndrome (DRRS, OMIM 607323), a rare autosomal dominant condition characterized by radial ray defects and Duane anomaly (a form of strabismus). Other features of Okihiro/ DRRS patients are anal, renal, cardiac, ear, and foot malformations, hearing loss, postnatal growth retardation, and facial asymmetry. The SALL4 gene product is a zinc ﬁnger protein thought to act as a transcription factor. It contains three highly conserved C2H2 double zinc ﬁnger domains, which are evenly distributed. A single C2H2 motif is attached to the second domain, and at the amino terminus SALL4 contains a C2HC motif. All but two reported SALL4 mutations lead to preterminal stop codons and are thought to cause the phenotype via haploinsuﬃciency.
The detection of larger deletions involving the whole SALL4 gene or single exons in patients with Okihiro or acro-renal-ocular syndrome provided proof for haploinsufﬁciency as the cause of the malformations. SALL4 is the only gene currently known to be mutated in patients with Okihiro/DRR syndrome. Mutations of SALL4 have been found in >90% of patients with classical Okihiro syndrome . Sall4 is regulated by Tbx5 in mouse and zebraﬁsh [7,8]. In the mouse, it controls Fgf10 expression in a synergistic manner together with Tbx5 in the forelimbs or with Tbx4 in the hindlimbs via direct eﬀects on the Fgf10 promoter, whereas the eﬀect of Sall4 and Tbx1 coexpression on the Fgf10 promoter is only additive . The cooperative action of Sall4 with Tbx5 or Tbx4 can be counteracted by Tbx2 and Tbx3. In the heart, mouse Sall4 interacts with Tbx5 and activates Gja5 but interferes with Tbx5-dependent activation of Nppa. In zebraﬁsh, fgf10 expression also depends on both tbx5 and sall4 . Here, fgf10 expression is activated by fgf24 via fgfr2. While tbx5 controls expression of fgf24, sall1a (SALL1 orthologue), and sall4, fgf10 expression is indirectly activated by sall1a and sall4 via activation of fgfr2 expression. Although these data provide information on one pathway for SALL4 regulation, the SALL4 promoter and its regulation has neither been described in the mouse nor in human or zebraﬁsh. Here we report on the cloning and functional analysis of the human SALL4 promoter.
Identiﬁcation of SALL4 expressing cell lines
SALL4 expression had previously been detected in human teratocarcinoma cell lines H12.1 and 2102 EP by Northern blotting . Since faint SALL4 expression had also been observed in human ovary tissue by RT-PCR, we analyzed SALL4 mRNA expression in human OVCAR-3 and OV-MZ-9 epithelial ovarian cancer cell lines by quantitative real-time RT-PCR. Only the 2102 EP cells as positive control and the epithelial ovarian cancer cells OVCAR-3 expressed SALL4 mRNA but not OVMZ-9 cells (Fig. 1). Interestingly, 2102 EP cells express SALL4 at an 18.7-fold higher level than OVCAR-3 cells and at levels similar to GAPDH.
Fig. 1. Real-time RT-PCR of human SALL4 mRNA from 2102 EP, OVCAR-3, and OV-MZ-9 cells. 2102 EP embryonal carcinoma cells show a 18.7-fold higher SALL4 expression level than epithelial ovarian cancer cells OVCAR-3, whereas human ovarian cancer cells OV-MZ-9 appeared to be consistently negative. Values are shown in comparison to endogenous GAPDH expression levels in the top left-hand corner of the diagram. Negative controls lacking template were always below threshold. The dissociation curve conﬁrmed the ampliﬁcation of only a single amplicon and veriﬁed the absence of secondary PCR products. Such single peak dissociation curves were found for all positive mRNAs in our samples.
Cloning of the 50 end of the SALL4 cDNA and identiﬁcation of additional SALL4 transcripts
In order to identify the transcriptional startpoint of SALL4,5 0 RACE was used. Two ampliﬁcation products of approximately 500 bp and 750 bp were obtained and sequenced. The 500 bp fragment contained parts of exon 2 (initial 130 base pairs) as well as exon 1 including the ATG and additional 28 base pairs in comparison to the SALL4 mRNA database sequence (NM020436), placing a transcriptional start site at 95 bp 50 of the ATG. This ﬁnding is supported by several ESTs (DA666635, DB066881, andCN308408) carrying similar 50 ends. No sequence identifying an additional upstream exon preceding exon 1 could be identiﬁed. The 750 bp fragment contained sequences of two putative alternative exons, positioned within intron 1 and following the AG-GT rule. One putative exon starts at position IVS1406 and ends at position IVS1574, covering 169 base pairs. The other putative exon starts at position IVS1+4577 and ends at position IVS1+4853, spanning 277 base pairs. These putative exons do not contain ATG start codons followed by an open reading frame, indicating the presence of an alternative transcript of SALL4 starting with a downstream ATG start codon within exon 2 (Fig. 2).
Fig. 2. Schematic representation of the SALL4 gene and the two detected transcript variants including alternative ﬁrst exons. (above) The most common variant represented by database sequence NM020436 contains exons 1, 2, 3, and 4. Translation starts with the ATG in exon 1. (below) Alternative transcript variant detected by 50 RACE. Transcription starts with exon 1a and includes exon 1b, both located within the large 50 intron. Zinc ﬁnger domains are represented by rhombuses within the boxes, grey color indicates untranslated regions. Since neither contains an in-frame ATG, translation is likely to start at the ﬁrst ATG 30 of the region coding for the ﬁrst (C2HC) zinc ﬁnger domain. Similar variants were detected in the SALL3 gene .
Cloning and transcriptional characterization of the SALL4 promoter
In order to clone the SALL4 promoter, 5 initial constructs overlapping at their 30 ends were generated by PCR ampliﬁcation of up to 1971 bp upstream of the SALL4 ATG start codon and cloning into the promoterless pGL3-Basic luciferase reporter plasmid. These constructs were transiently transfected into 2102 EP and OVCAR-3 cells, and co-transfected with pRL-SV40 vector in order to normalize luciferase activity in reference to the Renilla reniformis luciferase activity. Reporter constructs starting at 1971, 1446, 1003, 514, and 358 bp 50 of the ATG start codon demonstrated on average a 28-fold expression of luciferase (with no signiﬁcant diﬀerence among the diﬀerent constructs) in comparison to the empty pGL3-Basic vector in 2102 EP cells (Fig. 3A). Reporter gene activation was signiﬁcantly lower (on average 10-fold as compared to pGL3-Basic) in OVCAR3 cells than in 2102 EP cells (Fig. 3B). Murine and human putative promoter regions were aligned and showed 88% homology over 375 bp upstream of ATG (Fig. 4). More 50 sequences did not reveal any signiﬁcant homology. Further constructs of 278, 249, 218, 188, and 160 bp were generated and transfected to map the sequences responsible for transcriptional activation. The smallest construct showing full promoter activity included 249 bp 50 of ATG, while the constructs containing 218, 188, or 160 bp 50 of ATG did not show any signiﬁcant higher luciferase activity than the promoterless pGL3-Basic vector, indicating that important DNA sequences for driving SALL4 expression through binding of activating proteins reside within only 31 base pairs. Analysis of those 31 base pairs with the ‘‘genomatix Matinspector’’ (http://www.genomatix.de) revealed two highly conserved recognition sequences for HepG2-speciﬁc P450 2C factor-1 (ZNF83) and TCF/LEF, involved in the Wnt
signal transduction pathway. Additionally, a classical CAAT-box is positioned at 165 and two putative GATA-binding sites are positioned in direct proximity at 144 and 123 from ATG.
Fig. 3. Analysis of luciferase reporter constructs for minimal DNA sequence required for human SALL4 promoter activity. (A) Constructs were transiently transfected into 2102 EP cells and normalized to Renilla reniformis luciferase activity. Constructs containing at least 249 bp 50 upstream of ATG demonstrated functional activity, whereas a construct containing 218 bp 50 of ATG did not show any signiﬁcant luciferase activity. Reporter constructs comprising 1971, 1446, 1003, 514, and 358 bp did not show any signiﬁcant diﬀerence in activating luciferase expression. In average, a 28-fold increase in luciferase activity was observed compared to the empty pGL3-Basic vector. Each transfection was performed at least in triplicate using two diﬀerent DNA preparations of each construct. (B) Luciferase assay of transiently transfected constructs in OVCAR3 cells. As in 2102 EP cells, the minimal sequence stretch required for complete promoter activity contains 249 bp upstream of ATG. Smaller constructs do not promote any signiﬁcant reporter gene expression. Compared to the empty pGL3-Basic vector, a 7.9-fold uprating luciferase activity was ascertained in average.
The SALL4 promoter region is highly conserved in mammals
Comparison of 367 bp upstream of the ATG of human SALL4 with corresponding sequences of Pan trogloydes, Mus musculus, Canis familialis or Bos taurus (http://www.ebi.ac.uk/clustalw/index.html) revealed homologies between 89% (Mus musculus) and 99% (Pan trogloydes)(Fig. 4). For 2000 bp 50 of the translational start point (not shown), the homology between human and chimpanzee is 98%, but this ratio decreases to 75% in Canis familialis and Bos taurus, and mouse and human sequences correspond only weakly (58%), with highly conserved regions residing adjacent to exon 1. The TCF/LEF-binding motiv TACAAAG is fully conserved with the exception of the putative mouse Sall4 promoter. Here, the last adenine is altered to a guanine, but this does not change the binding speciﬁcity for LEF1 . The CAAT-box (165 from ATG) is identical between the analyzed species (Fig. 4).
Fig. 4. Cross-species sequence comparison between Homo sapiens, Pan trogloydes, Mus musculus, Canis familialis, and Bos taurus of the 50 upstream noncoding region of SALL4 and its orthologues. (A) Sequence identity to SALL4 varies between 89% (Mus musculus) and 99% (Pan trogloydes) within 367 base pairs upstream of ATG. (B) Sequence motifs putatively moderating SALL4 expression, accentuated by rectangles, as the CAAT-Box (165 from ATG) or GATA binding sites (144 and 123 from ATG) are highly conserved throughout the analyzed species. The TCF/LEF-binding motiv TACAAAG is invariable except in case of the putative mouse Sall4 promoter, where the third adenine is altered to a guanine. Asterisks indicate identical nucleotides at selected position across species. The position 1 denominates the ﬁrst nucleotide 50 of ATG.
Mutation of the TCF/LEF-binding site
Since the closely related Sall1 protein was found to enhance the canonical Wnt signaling pathway, we sought to test if LEF1 binding was crucial for SALL4 expression. The core-binding sequence of LEF1 was altered from TACAAAG to GCACCCT by site-directed mutagenesis in the 1997Luc construct. The luciferase activity of the mutated construct was strongly reduced to approximately 36% as compared with the activity of the wild type (Fig. 5). In comparison to the empty pGL3-Basic vector, the mutated construct showed a 11.2-fold luciferase expression, while the wild-type construct had a 31.0-fold increase of promoter activity.
Fig. 5. Transcriptional activity of the SALL4 promoter with an altered TCF/LEF-binding motif compared to wild type. Constructs were transiently transfected at least in triplicate into 2102 EP cells using standard procedures and incubated for 24 h. Values were calculated by normalizing against Renilla reniformis luciferase activity and compared to the empty pGL3-Basic vector. Luciferase activitiy of the mutated construct appeared to be reduced to 36% of the wild-type construct, showing a 11.2- and 31-fold increase of promoter activity, respectively.
pGL3 empty vector + – – – – – –
SALL4 promoter-Luc – + + + – – –
SALL4 promoter mut-Luc – – – – + + +
β-catenin + – + + + + +
LEF1 + – + – – + –
TCF4E – – – + – – +
Empty vector – + – – + – –
Rel. luciferase activity
Co-transfection of SALL4-luciferase constructs and TCF/ LEF
Since human embryonal kidney cells HEK293 are known to exhibit a low level of endogenous Wnt signaling, we selected those cells for transfections and subsequent reporter gene assays. SALL4-Luc constructs, either wild type or bearing a mutated binding motif for TCF/LEF, were co-transfected with expression constructs for b-catenin and LEF1 or TCF4E. As the expression vectors for LEF1 and TCF4E were not identical, expression levels of both proteins were analyzed by immunoblot using HA-antibody (BD Clontech, Heidelberg, Germany) in HEK293 of the SALL4 promoter and members of the TCF/LEF family in HEK293 cells. Induction is depicted relative to Renilla reniformis luciferase activity. Cotransfections were performed using equal amounts of DNA and adjusted concentrations of expression vectors previously analyzed by immunoblot. Promoter activity of SALL4 is enhanced by overexpressing LEF1 or TCF4E in concert with b-catenin. Transfection of the mutated construct resulted in a comparatively 68% lower induction of promoter activity. Overexpression of LEF/TCF4E promotes rescue of the mutated construct resulting in luciferase values comparable to single transfection of the wild-type SALL4 promoter in HEK293 cells transfected with these expression plasmids (data not shown). Band intensities were compared and adjusted amounts of expression plasmids were subsequently used for reporter gene assays (Fig. 6). In comparison to the promoterless pGL3-Basic vector co-transfected with b-catenin and LEF1 expression vectors, transfection of SALL4-Luc plasmid resulted in a 4.8-fold increase in reporter gene activity. 13.2-fold luciferase activity was achieved by co-transfection of SALL4-Luc with bcatenin and LEF1 expression vectors. Co-transfection of SALL4-Luc with b-catenin and TCF4E expression plasmids revealed 10.6-fold expression of luciferase. Co-transfection of SALL4 mut-Luc and the LEF1 or TCF4E and b-catenin expression vectors in a 1:30/1:6 ratio resulted in a weaker induction of promoter activity as compared to the wild-type construct (4.2-fold as compared to 10.6-fold normalized luciferase activity). No statistically signiﬁcant diﬀerence was seen between LEF1 and TCF4E. However, overexpression of these constructs in concert with b-catenin promotes rescue of the mutated SALL4 promoter leading to luciferase values comparable to transfection of the wild-type SALL4 promoter without LEF1/ TCF4E, indicating that altering the binding motif from TACAAAG to GCACCCT is not suﬃcient to abrogate aﬃnity under the experimental conditions of cell culture mediated luciferase assays.
Fig. 6. Co-transfection of the SALL4 promoter and members of the TCF/LEF family in HEK293 cells. Induction is depicted relative to Renilla reniformis luciferase activity. Cotransfections were performed using equal amounts of DNA and adjusted concentrations of expression vectors previously analyzed by immunoblot. Promoter activity of SALL4 is enhanced by overexpressing LEF1 or TCF4E in concert with b-catenin. Transfection of the mutated construct resulted in a comparatively 68% lower induction of promoter activity. Overexpression of LEF/TCF4E promotes rescue of the mutated construct resulting in luciferase values comparable to single transfection of the wild-type SALL4 promoter in HEK293 cells.
LEF1 binds to the SALL4 promoter in vitro
LEF 1 (Lef1) is expressed in high levels in pre-B and Tcells in adult mice, and in the neural crest, mesencephalon, tooth germs, whisker follicles, and other sites during mouse embryonic development . By Western blot analysis, we conﬁrmed that LEF1 is also expressed in human 2102 EP cells (data not shown). To determine whether LEF1 protein interacts with the TCF/LEF-binding motif within the SALL4 promoter in vitro, we performed gel retardation assays. Nuclear extracts from 2102 EP cells as well as recombinant LEF1 and TCF4E proteins were incubated with a [c-32P]dATP-labeled synthetic double-strand DNA probe of the SALL4 promoter with the normal and a mutated TCF/LEF-binding site in the middle. A single complex was observed when the SALL4 probe was incubated with rising amounts of recombinant LEF1 protein (Fig. 7).
Fig. 7. Electrophoretic mobility shift assay conﬁrms in vitro binding of LEF1 to a LEF/TCF consensus site within the SALL4 promoter region. (A) Recombinant LEF1 (100 and 300 ng) binds to a LEF/TCF consensus site in the SALL4 wild-type (WT) probe (lanes 2 and 3) and can be supershifted by the addition of a-LEF1 antibody (lane 4). It does not bind to a mutated (MUT) LEF/TCF site (lanes 6 and 7). As negative controls, no labeled probe was added in lanes 1 and 5. (B) LEF1 in nuclear extract derived from 2102 EP (2 and 6 lg) binds to SALL4 wild-type probe (lanes 2 and 3) and TCRa wt positive control probe (lanes 8 and 9) but not to a TCRa mutant probe (lanes 5 and 6). In lanes 1, 4, and 7 no radioactively labeled probe was added.
The application of a speciﬁc polyclonal rabbit a-LEF1 antibody resulted in a supershift of the complex. No retardation could be detected in case of the mutant probe. Adding recombinant TCF4E instead of LEF1 in a further assay also resulted in retardation of the wild-type probe (data not shown). However, members of the TCF family share highly conserved sequence homology and demonstrate similar aﬃnity to canonical-binding sites. Nuclear extract proteins from 2102 EP cells bound to labeled LEF1 probe but not to the mutated probe in presence of unspeciﬁc competitor. Mutant and wild-type probes derived from the promoter of nuclear T-cell receptor (TCR)a  served as a control since it is known to bind LEF1 protein.
In this report, we describe the ﬁrst characterization of the SALL4 promoter and the activation of SALL4 by members of the TCF/LEF family in the canonical WNT signaling pathway. In order to identify the most suitable cell line for reporter gene studies, SALL4 mRNA expression was analyzed by quantitative real-time RT-PCR. SALL4 was found to be expressed in OVCAR-3 cells, but not as strong as in 2102 EP cells. In OV-MZ-9 cells, no expression was detected. By 50 RACE, the transcriptional startpoint was mapped to 95 bp 50 of the ATG start codon, and two additional, yet undescribed, exons 50 to exon 2 were identiﬁed. Reporter constructs were generated containing up to 1977 bp 50 of the ATG. Transfection of those constructs into 2102 EP cells identiﬁed a region responsible for reporter gene activation under the conditions tested within 249 bp upstream of the ATG. Transfection of other constructs containing 218, 188, and 160 bp 50 of the ATG revealed no reporter gene activation, rendering a region of 31 bp being responsible for activation of reporter gene expression. The region contains a TCF/LEF-binding site, and mutation of this site resulted in decreased but not abolished activation of the SALL4 promoter. Co-transfection of SALL4-reporter gene constructs and LEF1/b-catenin or TCF4E/ b-catenin conﬁrmed SALL4 activation by LEF1 but also by TCF4E. EMSA studies provided evidence of direct binding of both members of the TCF/ LEF family to the presumed binding site, which was further substantiated by a supershift observed upon incubation with a LEF1 antibody.
Identiﬁcation of alternative transcripts
By 50 RACE, we could identify a previously unknown SALL4 transcript containing exons 1a and 1b instead of exon 1. No transcript was detected containing exon 1 in addition to exons 1a or 1b, suggesting that the two detected variants are the only existing mRNAs with diﬀerent 50 ends. These exons are conserved in the genomic SALL4 sequence of Pan trogloydes but not in mouse Sall4 (data not shown). Interestingly, translation of the exon 1a–1b transcript would likely result in a SALL4 protein not containing the conserved amino terminal region including the C2HC zinc ﬁnger domain and the region which contains the amino terminal repressor domain in SALL1 [12,13]. A similar transcript had previously been detected for the gene SALL3, which appeared to be the most closely related SALL gene [2,3]. The resulting protein would also miss a conserved stretch of 12 amino acids encoded by exon 1 important for recruitment of the nucleosome remodeling and deacetylase corepressor complex (NuRD) involved in global transcriptional repression and regulation of speciﬁc developmental processes . Although SALL4 and SALL3 are not (yet) known to be transcriptional repressors as shown for SALL1, the existence of a transcript lacking essential repressor domains might indicate that the alternative transcripts in SALL4 and SALL3 reﬂect diﬀerent functional properties.
Structure of the SALL4 promoter
By 50 RACE, the transcriptional start-point was mapped to 95 from ATG. No consensus TATA box was found within 2 kb upstream of this site, suggesting that SALL4 has a TATA-less promoter as shown for SALL1 and 2 [15,16]. Only one putative, but non-canonical, TATA box in proximity to the transcriptional start site was detected, TAATAATAATTA, 41 nt from the predicted initiator region (INR). The consensus for the INR in mammalian cells is Py-Py(C)-A+1-N-T/A-Py-Py, ﬁtting quite well with the predicted INR element for SALL4 (CCCAACTCC, Fig. 4). INR regions are important for binding the basal transcription factor TFIID in a sequence speciﬁc manner and cooperatively with a DPE motif. The DPE is found most commonly in TATA-less promoters and is located at +28 to +32 relative to the A+1 position in the INR with a consensus sequence A/G+28-G-A/T-C/T-G/A/C (for review, see Ref. [33–35]. Such a sequence cannot be found within the SALL4 promoter, indicating that this promoter may be TATA-box dependent despite the lack of a classical TATA box. On the other hand, TATA-less promoters are often associated with multiple transcription start sites.
The results of the 50 RACE in comparison with the sequences of several ESTs suggest multiple transcription start sites for SALL4, since the transcription start site obviously varies in a few nucleotides. By reporter gene analysis, we could demonstrate that a minimal region of 31 bp is important for SALL4 expression in the 2102 EP cell line. The region contained a TCF/LEF-binding site highly conserved throughout mammals, which we could show to be functional by reporter studies and mobility shift assays. Earlier reports described activation of Sall4 in mouse and sall4 in zebraﬁsh by Tbx5 , but no Tbx5-binding site was detected within 2 kb upstream of the ATG in exon 1 of SALL4. This does not exclude a direct interaction of SALL4 by TBX5 but indicates that TBX5 might rather bind to a limb-speciﬁc enhancer. Within the 31 bp region crucial for SALL4 expression, only one other binding site for ZNF83 was predicted, which however overlaps with the TCF/LEF-binding site. Mutation of the TCF/LEF-binding site in the 1997 bp construct resulted in a decreased but not abolished activation of the SALL4 promoter, which might be explained by other activating factors binding upstream of the TCF/LEF site.
SALL4 is a target of the canonical WNT signaling
Our results provide a new and important aspect on the relation of SALL genes and WNT signaling. Studies in chicken have shown that ectopic expression of the SALL1 homologue csal1 is depending upon interaction of wnt3a or wnt7a with Fgf4 . Studies in mouse and Xenopus laevis have shown that SALL genes can also actively act on the Wnt signaling pathway. While Sall1 was found to interact with β-catenin and to enhance the canonical Wnt signaling by binding to heterochromatin , the Xsal2 protein acts by repressing canonical Wnt signaling .
Lymphoid enhancer-binding factor-1 (LEF-1) belongs to the high mobility group (HMG) regulating, e.g., genes involved in T-cell development through binding the core consensus sequence 50-CTTTGA/TA/T-3 in the minor groove of the DNA [20, 21]. Binding induces formation of a sharp bend in the double helix, thus facilitating the binding of transcription factors to the adjacent DNA sequences required for driving transcription . TCF/LEF apparently operates as a repressor when not linked to b-catenin by interacting with Groucho, in turn recruiting histone deacetylases. These HDAC remove an acetyl group from histones, which allows histones to bind DNA and inhibit gene transcription [22–24]. Binding of β-catenin possibly replaces Groucho by the histone acetylase CBP/p300 (cyclic AMP response element-binding protein), inducing chromatin remodeling resulting in transcriptional activation of the target gene [25,26].
As in the cytoplasm and the extracellular compartment, negative regulators of the Wnt induced pathway also exist in the nucleus. Chibby and ICAT can bind β-catenin and thereby prevent its interaction with LEF1 [26,27]. Phosphorylation by MAPK related protein kinase also serves as a mechanism for regulating LEF1 activity, leading to a reduced capability of the LEF1/b-catenin complex to form a ternary complex with DNA [28, 29]. A key protein within the signaling pathway initiated by WNTs is β-catenin. The primary structure of the β-catenin protein comprises an N-terminal domain that is subject to phosphorylation events leading to its proteolytic degradation, a 42 amino acid arm repeat interacting with TCF/ LEF, and a C-terminal domain important for transcriptional activation.
Phosphorylation, and hence targeting of β-catenin for ubiquitination and degradation, through glycogen synthase kinase-3 β (GSK3) is faciliated by Axin and Adenomatous Polyposis Coli (APC), constituting a kind of scaﬀold. Cytoplamic β-catenin levels are consequently kept low. Binding of the signal molecule Wnt to its transmembrane receptor Frizzled induces diverse processes leading to an inhibition of the GSK3 activity and thus to a cytoplasmic accumulation of β-catenin. Free cytoplasmic β-catenin enters the nucleus and induces gene expression by complexing TCF/LEF and forming a ternary complex with DNA (for review, see Ref. ).
In addition to enhancement or repression of the canonical WNT pathway by SALL genes [18,19], we report here a further interaction of a SALL gene with the canonical WNT pathway by demonstrating that the canonical WNT signaling can directly regulate expression of SALL4, and that this regulation is achieved by direct interaction of LEF1 and the SALL4 promoter. A large number of Wnt targets have been identiﬁed, including also members of the Wnt signaling pathway resulting in feedback regulation and underlying the complex machinery working in concert for ﬁne-tuning signal transduction. One of the key roles of Wnt signaling concerns cell proliferation, and lack of Wnt5a inhibits outgrowth of the limbs , a major phenotype seen in patients with Okihiro syndrome. Mutation of human WNT3 results in tetraamelia , and mutation of WNT7A in Fuhrmann and Al-Awadi/Raas-Rothschild/Schinzel phocomelia syndromes , supporting an important role of WNT signaling in the pathogenesis of human limb malformations. The ﬁnding of a direct regulation of SALL4 by TCF/LEF supports now the assumption that SALL4 may be activated by WNT3 and/ or WNT7A by means of TCF/ LEF during limb development.
 R. Al-Baradie, K. Yamada, C. St Hilaire, W.M. Chan, C. Andrews, N. McIntosh, M. Nakano, E.J. Martonyi, W.R. Raymond, S. Okumura, M.M. Okihiro, E.C. Engle, Duane Radial Ray Syndrome (Okihiro syndrome) maps to 20q13 and results from mutations in SALL4, a new member of the SAL family, Am. J. Hum. Genet. 71 (2002) 1195–1199.  J. Kohlhase, S. Hausmann, G. Stojmenovic, C. Dixkens, K. Bink, W. Schulz-Schaeﬀer, M. Altmann, W. Engel, SALL3, a new member of the human spalt-like gene family, maps to 18q23, Genomics 62 (1999) 216–222.  J. Kohlhase, M. Heinrich, L. Schubert, M. Liebers, A. Kispert, F. Laccone, P. Turnpenny, R.M. Winter, W. Reardon, Okihiro syndrome is caused by SALL4 mutations, Hum. Mol. Genet. 11 (2002) 2979–2987.  J. Kohlhase, A. Ko ¨hler, H. Ja ¨ckle, W. Engel, R. Stick, Molecular cloning of a SALL1-related pseudogene and mapping to chromosome Xp11.2, Cytogenet. Cell Genet. 84 (1999) 31–34.  J. Kohlhase, R. Schuh, G. Dowe, R.P. Ku ¨hnlein, H. Ja ¨ckle, B. Schroeder, W. Schulz-Schaeﬀer, H.A. Kretzschmar, A. Ko ¨hler, U. Mu ¨ller, M. Raab-Vetter, E. Burkhardt, W. Engel, R. Stick, Isolation, characterization, and organ-speciﬁc expression of two novel human zinc ﬁnger genes related to the Drosophila gene spalt, Genomics 38 (1996) 291–298.  J. Kohlhase, D. Chitayat, D. Kotzot, S. Ceylaner, U.G. Froster, S. Fuchs, T. Montgomery, B. Rosler, SALL4 mutations in Okihiro syndrome (Duane-radial ray syndrome), acro-renal-ocular syndrome, and related disorders, Hum. Mutat. 26 (2005) 176–183.  S.A. Harvey, M.P. Logan, sall4 acts downstream of tbx5 and is required for pectoral ﬁn outgrowth, Development 133 (2006) 1165–1173.  K. Koshiba-Takeuchi, J.K. Takeuchi, E.P. Arruda, I.S. Kathiriya, R. Mo, C.C. Hui, D. Srivastava, B.G. Bruneau, Cooperative and antagonistic interactions between Sall4 and Tbx5 pattern the mouse limb and heart, Nat. Genet. 38 (2006) 175–183.  J.J. Love, X. Li, D.A. Case, K. Giese, R. Grosschedl, P.E. Wright, Structural basis for DNA bending by the architectural transcription factor LEF-1, Nature 376 (1995) 791–795.  M. Oosterwegel, M. van de Wetering, J. Timmerman, A. Kruisbeek, O. Destree, F. Meijlink, H. Clevers, Diﬀerential expression of the HMG box factors TCF-1 and LEF-1 during murine embryogenesis, Development 118 (1993) 439–448.  J. Galceran, S.C. Hsu, R. Grosschedl, Rescue of a Wnt mutation by an activated form of LEF-1: regulation of maintenance but not initiation of Brachyury expression, Proc. Natl. Acad. Sci. USA 98 (2001) 8668–8673.  S. McLeskey Kiefer, K.K. Ohlemiller, J. Yang, B.W. McDill, J. Kohlhase, M. Rauchman, Expression of a truncated Sall1 transcriptional repressor is responsible for Townes-Brocks syndrome birth defects, Hum. Mol. Genet. 12 (2003) 2221–2227.  C. Netzer, S. Bohlander, M. Hinzke, Y. Chen, J. Kohlhase, Deﬁning the heterochromatin localization and repression domains of SALL1, Biochim. Biophys. Acta 1762 (2006) 386–391.  S.M. Lauberth, M. Rauchman, A conserved twelve amino acid motif in sall1 recruits nuRD, J. Biol. Chem. (2006).  L. Chai, J. Yang, C. Di, W. Cui, R. Lai, Y. Ma, Transcriptional activation of the sall1 by the human six1 homeodomain during kidney development, J. Biol. Chem. 281 (2006) 18918–18926.  Y. Ma, D. Li, L. Chai, A. Luciani, D. Ford, J. Morgan, A. Maizel, Cloning and characterization of two promoters for the human HSAL2 gene and their transcriptional repression by the Wilms tumor suppressor gene product, J. Biol. Chem. 276 (2001) 48223–48230.  E.R. Farrell, A.E. Munsterberg, csal1 is controlled by a combination of FGF and Wnt signals in developing limb buds, Dev. Biol. 225 (2000) 447–458.  A. Sato, S. Kishida, T. Tanaka, A. Kikuchi, T. Kodama, M. Asashima, R. Nishinakamura, Sall1, a causative gene for Townes-Brocks syndrome, enhances the canonical Wnt signaling by localizing to heterochromatin, Biochem. Biophys. Res. Commun. 319 (2004) 103–113.  T. Onai, N. Sasai, M. Matsui, Y. Sasai, Xenopus XsalF: anterior neuroectodermal speciﬁcation by attenuating cellular responsiveness to Wnt signaling, Dev. Cell 7 (2004) 95–106.
7.11.5 SALL4. An emerging cancer biomarker and target
SALL4 is a transcription factor that plays essential roles in maintaining self-renewal and pluripotency of embryonic stem cells (ESCs). In fully differentiated cells, SALL4 expression is down-regulated or silenced. Accumulating evidence suggest that SALL4 expression is reactivated in cancer. Constitutive expression of SALL4 transgene readily induces acute myeloid leukemia (AML) development in mice. Gain- and loss-of-function studies reveal that SALL4 regulates proliferation, apoptosis, invasive migration, chemoresistance, and the maintenance of cancer stem cells (CSCs). SALL4 controls the expression of its downstream genes through both genetic and epigenetic mechanisms. High level of SALL4 expression is detected in cancer patients, which predicts adverse progression and poor outcome. Moreover, targeted inhibition of SALL4 has shown efficient therapeutic effects on cancer. We have summarized the recent advances in the biology of SALL4 with a focus on its role in cancer. Further study of the oncogenic functions of SALL4 and the underlying molecular mechanisms will shed light on cancer biology and provide new implications for cancer diagnostics and therapy.
SALL4 (sal-like4) is a member of the mammal homologs of Drosophila homeoticgene spalt (sal). In humans, SALL4 gene mutations are known to cause Okihiro syndrome (Duane radial ray syndrome), an autosomal dominant disease involving multiple organ defects [1–4]. In mice, SALL4 homozygous knockout is embryonic lethal and SALL4 heterozygous knockout causes dysplasia of multiple organs [5,6]. SALL4 is an essential factor for the maintenance of self-renewal and pluripotency of embryonic stem cells(ESCs)[7,8]. SALL4 expression gradually decreases with the maturation of tissues and organs. However, SALL4 expression is found to be restored in numerous human malignancies. High levels of SALL4 has been observed in both hematological diseases and solid tumors, including acute myeloid leukemia, chronic myeloid leukemia, breast cancer, lung cancer, gastric cancer, colorectal cancer, liver cancer, endometrial cancer and glioma. SALL4 acts as an oncogene that plays multifaceted roles in the processes of cancer initiation, development, and progression. Exploring the underlying mechanisms responsible for the oncogenic functions of SALL4 will allow the development of a novel target for cancer therapy. In this review, we focus on recent progress in understanding the roles of SALL4 in cancer and the molecular mechanisms, with an emphasis on the potential of SALL4 in cancer diagnostics and treatment.
Roles of SALL4 in stem cells
SALL4 has two isoforms,S ALL4A and SALL4B, which resulted from different internal splicing patterns in exon 2 (Fig.1).
Fig. 1. SALL4 gene and protein structure. Human SALL4 gene localizes on chromosome 20q13.13-q13.2 and consists of four exons. The human SALL4 protein has multiple zinc ﬁnger (ZF) domains of the SAL type, which is composed of one N terminal C2HC type zinc ﬁnger (NT ZF) and seven C2H2 type zinc ﬁngers. A Q rich motif is responsible for interactions between members of SALL1-4. The SALL4 protein has two isoforms, SALL4A and SALL4B, which resulted from different internal splicing patterns in exon 2. SALL4B lacks the 386 to 822 amino acids of full-length SALL4 protein. A conserved “KRLR” motif at amino acid positions of 64–68 is identiﬁed in both SALL4A and SALL4B as nuclear localization signal (NLS).
SALL4A and SALL4B are able to form homodimers or heterodimers with distinct DNA binding sites and exhibit different roles during early embryogenesis . In murine ESCs, depletion of both isoforms of SALL4 by shRNA leads to multilineage differentiation. SALL4A and SALL4B have overlapping, but not identical binding sites of epigenetic marks in target loci or their interactions with other pluripotent factors. In addition, SALL4B, but not SALL4A, alone can maintain the pluripotent state of mouse ESCs. SALL4 expression could be detected in embryos as early as at the 4-cell stage and is gradually restricted to inner cell mass from which ESCs are normally derived. Disruption of both SALL4 alleles cause embryonic lethality during peri-implantation and depletion of SALL4 results in early embryonic development defects, suggesting that SALL4 is critical for early embryonic development. SALL4 plays a vital role in stem cell self-renewal and pluripotency through multiple layers of mechanisms.
First, SALL4 regulates the activation of several important signaling pathways in stem cells. Activation of Wnt/β-catenin signaling maintains the pluripotency of human and mouse embryonic stem cells . SALL4 binds to β-catenin and up-regulates the expression of the target genes of the Wnt/β-catenin pathway , suggesting that SALL4 may promote stem cell self-renewal and inhibit stem cell differentiation through the interaction with β-catenin. STAT3 activation mediates the self-renewal and pluripotency of embryonic stem cells . SALL4 may interact with STAT3 to regulate the self-renewal and differentiation of stem cells. The Hedgehog signaling pathway plays a pivotal role in organogenesis and differentiation during development . Genome-wide analysis reveals that SALL4 regulates Sonic Hedgehog (SHH) pathway . SALL4 may modulate SHH signaling to prevent the differentiation of embryonic stem cells. SALL4 inhibits the expression of PTEN and induces the activation of the Akt pathway , which may enhance stem cell self-renewal and expansion and maintain stem cells at undifferentiated state.
Second, SALL4 modulates the transcription of key stemness factors including Oct4, Nanog, Sox2, and c-Myc [7,15–17]. Compared to wild type ESCs, the expression of these 4 genes is remarkably down-regulated in SALL4+/− ESCs , suggesting that SALL4 could be a key regulator for stem cell factors. SALL4 can activate the expression of Oct4, Nanog, Sox2, and c-Myc and form a transcriptional core network with these factors to maintain cell stemness. The down-regulation of core stemness factors may be responsible for SALL4 knockdown-induced spontaneous cell differentiation . Due to its critical role in maintaining pluripotency, SALL4 has been used as an enhancer for induced pluripotent stem cell (iPS) generation from somatic cells .
Third, SALL4 may regulate the expression of key genes that are associated with stem cell self-renewal and differentiation through epigenetic modulation. SALL4 induces the activation of Bmi-1, an important factor for regulating stem cell self-renewal, by mediating H3K4 trimethylation and H3K79 dimethylation at thepromoter region . In addition, SALL4 recruits MLL (mixed lineage leukemia), a histone methyltransferase, to prompt H3K4 and H3K79 methylation, resulting in HOXA9 up-regulation .
In summary, these ﬁndings indicate that SALL4 is involved in regulating self-renewal and pluripotency of stem cells through a variety of signaling pathways, transcription factors, and epigenetic modulators. SALL4 expression has also been found in adult stem/progenitor cells. In human bone marrow, SALL4 expression is strictly limited to the CD34+ hematopoietic stem/progenitor cells (HSCs/HPCs) and decreased following differentiation [22,23]. SALL4 is found to be a robust stimulator for both human and mouse HSC/HPC self-renewal [24,25]. Human HSCs transduced with SALL4 are able to expand rapidly and eﬃciently in vitro. On the contrary, depletion of endogenous SALL4 leads to reduced HSC proliferation and accelerated cell differentiation. SALL4 regulates the expression of genes that are critical in maintaining short-term and long-term HSC proliferation, including Bmi-1, HOXA9, and c-Myc . SALL4 works with these factors to form a concerted network for normal hematopoiesis . In addition to HSCs/HPCs, SALL4 is expressed in fetal hepatoblasts but silenced in adult hepatocytes . The expression levels of SALL4 gradually fall during liver development. SALL4 overexpression enhances while SALL4 knockdown impairs induced differentiation of hepatoblasts to cholangiocytes and the formation of bile duct, suggesting that SALL4 regulates cell fate decision in fetal hepatic stem/progenitor cells.
Roles of SALL4 in cancer
SALL4 is overexpressed in cancer and affects multiple cellular processes which are involved in tumorigenesis, tumor growth and tumor progression. SALL4 regulates a variety of targets in distinct types of cancer cells. In this section, we review the targets of SALL4 and their functions in cancer (Fig.2).
Fig. 2. Proposed model for SALL4 regulatory network in cancer. SALL4 regulates cell proliferation, apoptosis, migration/invasion, drug resistance, and stemness by targeting a variety of genes. SALL4 regulates cell proliferation through the β-catenin/cyclin D1, Bmi-1, and PTEN pathways. SALL4 regulates cell migration/invasion through the ZEB1/ E-cadherin and the c-Myc pathways. SALL4 inhibits apoptosis through the Bmi-1, HOXA9, and FADD pathways. SALL4 regulates the resistance of cancer cells to chemotherapy by targeting the ABCA3, ABCG2, and c-Myc pathways. SALL4 regulates the self-renewal of cancer stem cells through the Oct4, Nanog, Sox-2, and Bmi-1 pathways. STAT3, CDX1, and TCF/LEF are upstream positive regulators of SALL4. SALL4 is a downstream target of microRNA-107. Natural compounds matrine and apigenin could inhibit the expression of SALL4.
SALL4 and cell transformation
During normal hematopoiesis, SALL4 is expressed in the CD34+ HSC/HPC population. SALL4 expression is down-regulated or silenced in mature blood cells. In contrast, SALL4 is constitutively expressed in human primary acute myeloid leukemia (AML) and myeloid leukemia cell lines. To test whether constitutive expression of SALL4 is suﬃcient to induce AML, Ma and colleagues generated a SALL4B transgenic mouse model with SALL4B expression in most organs. They demonstrated that all the monitored SALL4B transgenic mice exhibit dysregulated hematopoiesis and develop myelodysplastic syndrome (MDS)-like features at ages 6–8 months and half of the monitored mice eventually progressed to AML . Mice injected with serially transplanted SALL4B-induced AML cells also develop aggressive AML, suggesting that SALL4B-induced AML is transplantable.
The potential signaling pathway that SALL4 may affect in leukemogenesis has been postulated, which includes SALL4 binding to β-catenin and activating the Wnt/β-catenin signaling pathway. The expression of Wnt/β-catenin downstream target genes, such as c-Myc and Cyclin D1, is upregulated in SALL4B transgenic mice. Thus, constitutive expression of SALL4 in AML may enable leukemic blasts to gain stem cell properties, such as self-renewal and/or lack of differentiation, and thus become leukemic stem cells (LSCs). In addition, Bmi-1 is identiﬁed as a target gene for SALL4 in both hematopoietic and leukemic cells .
Bmi-1 is a putative oncogene that modulates stem cell pluripotency and plays a role in leukemogenesis. SALL4 binds to Bmi-1 promoter and directly affects the levels of endogenous Bmi-1 expression. In vitro knockdown of SALL4 by siRNA in leukemic cells or in vivo deletion of one copy of SALL4 in mouse bone marrow signiﬁcantly reduces Bmi-1 expression. Bmi-1 expression is up-regulated in transgenic mice that constitutively overexpress SALL4B, and the levels of Bmi-1 in these mice increase as they progress from normal to preleukemic (myelodysplastic syndrome) and leukemic (acute myeloid leukemia) stages. Furthermore, there is a strong positive correlation between SALL4 and Bmi-1 expression in human AML samples. SALL4 induces high levels of H3K4 trimethylation and H3K79 dimethylation in the binding region of the Bmi-1 promoter, suggesting that SALL4 provides epigenetic modiﬁcations at the Bmi-1 gene promoter. These ﬁndings indicate a link between SALL4 and leukemogenesis by regulating self-renewal of leukemic stem cells.
SALL4 and cell proliferation and apoptosis
SALL4 acts as a key regulator of cell proliferation and apoptosis in cancer cells. SALL4 knockdown induces massive apoptosis and signiﬁcant growth arrest in human leukemic cells . In addition, reduction of SALL4 markedly diminishes tumorigenicity of human leukemia cells in immunodeﬁcient mice. ChIP-chip assay for the global gene target of SALL4 in human leukemic cells reveals that SALL4 binds to the promoter of genes that are critically involved in apoptosis. The expression levels of these genes change signiﬁcantly after SALL4 knockdown, indicating that SALL4 is a key regulator of apoptosis-associated genes. In addition, SALL4 has an important role in the proliferation and survival of chronic myeloid leukemia (CML) cells, and its expression is associated with an advanced stage of CML disease. Downregulation of SALL4 leads to cell cycle arrest and apoptosis in CML cells .
In AML and CML cells, SALL4 knockdown-induced apoptosis and cell cycle arrest are rescued by forced expression of Bmi-1, suggesting that SALL4 regulation of Bmi-1 may at least be partially responsible for its effects on cell proliferation and apoptosis. Moreover, HOXA9 is identiﬁed as another downstream target of SALL4 . SALL4 interacts with mixed lineage leukemia (MLL) and co-occupies the HOXA9 promoter region in AML leukemic cells. Compared with wild-type controls, HOXA9 is up-regulated in SALL4B transgenic mice. In primary human AML cells, downregulation of SALL4 also reduces HOXA9 expression and induces cell apoptosis. Furthermore, SALL4 knockdown leads to growth inhibition of lung cancer cells as a result of cell cycle arrest . Similarly, reduction of SALL4 expression by siRNA completely also inhibits the proliferation of breast cancer cells as a result of cell cycle arrest . Conversely, SALL4-overexpressing liver cancer cells exhibit enhanced cell proliferation with the characteristics of reduced cell population in the G1 phase through the up-regulation of cyclin D1 and D2 . In contrast, down-regulation of SALL4 inhibits liver cancer cell proliferation in vitro as well as in tumor xenograft models. We have recently reported that SALL4 overexpression enhances while knockdown of SALL4 inhibits the proliferation of gastric cancer cells . In consistent with this observation, SALL4 knockdown also inhibits endometrial cancer cell growth in vitro and tumorigenicity in vivo due to the inhibition of cell proliferation and increased apoptosis .
SALL4 and invasive migration
The research from our group has indicated that the SALL4 level is highly correlated with lymph node metastasis of gastric cancer . Enforced expression of SALL4 enhances the migration of human gastric cancer cells, whereas knockdown of SALL4 by siRNA leads to the opposite effects. SALL4 overexpression up-regulates the expression of Twist1, N-cadherin while down-regulating E-cadherin expression, thus inducing epithelial–mesenchymal transition (EMT) in gastric cancer cells. In endometrial cancer, down-regulation of SALL4 signiﬁcantly impedes the migration and invasion properties of cancer cells in vitro and their metastatic potential in vivo. SALL4 speciﬁcally binds to the c-Myc promoter region in endometrial cancer cells. Reduction of SALL4 leads to a decreased expression of c-Myc and ectopic SALL4 overexpression causes increased c-Myc expression, indicating that c-Myc is one of the SALL4 downstream targets in endometrial cancer. In addition, SALL4 suppresses E-cadherin expression and maintains cell dispersion in basal-like breast cancer . SALL4 inhibits intercellular adhesion and maintains cell motility after cell–cell interaction and cell division, which results in the dispersed phenotype. SALL4 knockdown leads to EMT in basal-like breast cancer cells. Further study showed that SALL4 positively regulates the EMT factor ZEB1, therefore suppressing E-cadherin transcription and leading to cell dispersion and mesenchymal gene expression.
SALL4 and cancer stem cells
SALL4 is essential for maintaining the properties of cancer stem cells (CSCs). The expression of SALL4 is signiﬁcantly higher in side population (SP) cells than that in non-SP cells in leukemia cell lines, suggesting that SALL4 is more abundant in CSCs . Knockdown of SALL4 leads to reduced frequency of SP cells, indicating that SALL4 is required for the self-renewal of cancer stem cells. Similarly, SALL4 is also enriched in SP of breast cancer cells and increased SALL4 expression leads to an expansion of SP subset in breast cancer cells.
We have recently demonstrated that SALL4 overexpression induces the acquirement of stemness in gastric cancer cells through increasing the levels of Sox2, Bmi-1, CD133, and Lin28B . SALL4 overexpressing gastric cancer cells could be eﬃciently induced to differentiate into osteoblasts and adipocytes under the appropriate conditions, suggesting that SALL4 overexpression endows gastric cancer cells with stemness and pluripotency. SALL4 is suggested as a stem cell marker in liver cancer that regulates the stemness of liver cancer cells . SALL4 overexpression down-regulates differentiation markers ALB, TTR, and UGT2B7, suggesting that SALL4 inhibits hepatocytic differentiation in human liver cancer cells. SALL4 is identiﬁed as one of the transcription factors that are potentially activated in hepatic stem cell-like HCC (HpSC-HCC). SALL4-positive HCCs are associated with expression of the hepatic stem cell markers including EpCAM.
EpCAM+ cells have higher expression of SALL4 and possess a stronger spheroid formation capacity than EpCAM− cells, indicating that SALL4 is activated in EpCAM+liver CSCs. Ectopic expression of SALL4 leads to up-regulation of the hepatic stem cell markers and down-regulation of the mature hepatocyte marker. Moreover, SALL4 overexpression results in the signiﬁcant activation of spheroid formation while knockdown of SALL4 results in a compromised spheroid formation capacity with decreased expression of EpCAM, suggesting that SALL4 regulates stemness of HpSC-HCC.
SALL4 and chemoresistance
SALL4 expression is associated with therapy response in cancer. In acute myeloid leukemia (AML), SALL4 expression is higher in drug resistant patients than those from drug-responsive cases. AML cells with decreased SALL4 expression are more sensitive to drug treatments than their parental cells. SALL4 modulates drug sensitivity through the maintenance of SP cells in AML . SALL4 directly binds to the promoter region of ABCA3, a resistance-mediating transporter, affecting the formation of SP cells in AML. SALL4 expression is positively correlated to that of ABCA3 in primary leukemic patient samples. In addition to AML patients, constitutive expression of SALL4 has also been observed in CML cells in the blast crisis or accelerated phase. Exposure to tyrosine kinase inhibitors (TKI) leads to increased expression of SALL4 in CML cells, which consequently upregulates ABCA3 . High ABCA3 levels facilitate detoxiﬁcation of TKI, protecting CML cells against TKI effects. The positive regulation of ABCA3 through SALL4 constitutes an auto-protective loop to protect CML cells from the lytic activity of TKI. Suppression of SALL4 by siRNA partially abrogates a TKI-associated increase in ABCA3 expression and increases susceptibility of CML cells to cytotoxicity of tyrosine kinase inhibition. Treatment with indomethacin interrupts the inducible SALL4/ABCA3 pathway in CML cells to restore TKI susceptibility.
Constitutive expression of SALL4 affects the sensitivity of endometrial cancer cells to carboplatin treatment . Overexpression of SALL4 in carboplatin sensitive endometrial cancer cells could further promote carboplatin resistance in a dose-dependent manner. SALL4-transfected endometrial cancer cells show increased proliferationaftercarboplatintreatmentcomparedwithcontrolcellswhile the overexpression also protects endometrial cancer cells from carboplatin-inducedapoptosis.Incontrast,incarboplatin-resistant endometrial cancer cells, SALL4 knockdown signiﬁcantly sensitizes the cells to carboplatin treatment. SALL4 directly regulates c-Myctranscriptionalactivityinendometrialcancercells,whichmay be partially responsible for the chemotherapy resistance induced by SALL4 upregulation. In addition, SALL4 expression is correlated with chemosensitivity in liver cancer cells. SALL4-overexpression induces survival and proliferation of liver cancer cells in response to 5-FU treatment, suggesting that SALL4 expression results in selection of chemoresistant cells .
SALL4 and epigenetic modulation
In addition to transcriptional control, SALL4 also regulates gene expression through epigenetic mechanisms. DNA methylation, histone modiﬁcation, chromatin remodeling, and non-coding RNAs are the four major molecular mechanisms responsible for epigenetic modiﬁcation. Yang et al. suggest that SALL4 protein directly interacts with DNA methyltransferases (DNMTs), indicating that SALL4 is able to repress transcription through recruitment of DNA methyltransferases . In addition, SALL4 has been shown to co-occupy target genes with polycomb repressive complex (PRC) .
SALL4 may repress gene transcription though the induction of PRC components (such as Bmi-1) or interaction with PRC complex members. Moreover, SALL4 interacts with histone lysine-speciﬁc demethylase1 (LSD1) to repress gene transcription in stem cells . In addition to gene repression, SALL4 is capable of binding to the histone methyltransferase MLL to activate HOXA9 gene expression , suggesting that SALL4-mediated methylation and demethylation in DNA and histone may distinctly regulate gene expression in stem cells and cancer.
Second, SALL4 is associated with Mi-2/nucleosome remodeling and deacetylase (NuRD) complex and theSALL4-interacting protein complex exhibits histone deacetylase (HDAC)activity . For instance, SALL4 co-occupies the same promoter regions of PTEN as HDAC2 and represses its expression in vitro, indicating that SALL4-repressed gene transcription could be mediated by histone deacetylation and nucleosome remodeling. However, there is a lack of studies about the regulatory effects of SALL4 on non-coding RNA in epigenetic modulation. Therefore, the existing data suggest that SALL4 may recruit multiple epigenetic modiﬁers to synergistically remodel local chromatin structure and coordinately regulate gene transcription (Fig.3).
Fig. 3. SALL4 and epigenetic machinery. SALL4 represses or activates gene transcription through the interaction with distinct epigenetic modiﬁers. SALL4 suppresses gene transcription through recruitment of DNA methyltransferases (DNMT), Mi-2/nucleosome remodeling and deacetylase (NuRD) complex, polycomb repressive complex (PRC), and histone demethylase (LSD1). SALL4 activates gene expression through recruitment of histone methyltransferase such as MLL (mixed-lineage leukemia).
SALL4 regulation in cancer
SALL4 is regulated at multiple levels in cancer. Aberrant hypomethylation of the promoter region of SALL4 gene is observed in MDS patients and SALL4 mRNA level is highly associated with the status of SALL4 hypomethylation, indicating that SALL4 gene expression is dysregulated in MDS patients by epigenetic mechanism . The frequency of SALL4 hypomethylation is signiﬁcantly increased in higher risk MDS patients, suggesting that hypomethylation of SALL4 gene is involved in the progression of MDS. In addition, SALL4 gene is aberrantly hypomethylated in acute myeloid leukemia patients and the status of SALL4 gene methylation is associated with intermediate and poor karyotypes of AML . SALL4 is also regulated by a variety of transcription factors that are closely linked with tumor development and progression.
Multiple STAT3-binding sites have been identiﬁed in the SALL4 gene promoter region. Down-regulation of STAT3 activity remarkably decreased the expression of SALL4 . STAT3-mediated SALL4 regulation is critical for the survival of breast cancer cells. SALL4 expression is also regulated by the canonical Wnt signaling pathway. SALL4 promoter contains a conserved TCF/LEF-binding site. Co-transfection of β-catenin with LEF1 (or TCF4E) greatly increases SALL4 luciferase activity while mutation of the TCF/LEF-binding site attenuates SALL4 luciferase activity, suggesting that SALL4 is a direct transcriptional target of canonical Wnt signaling . Furthermore, SALL4 interacts with β-catenin to cooperatively activate its target genes. Therefore, the regulation of SALL4 by Wnt signaling may form a feedback loop to ﬁne-tune the Wnt signaling pathway.
SALL4 is identiﬁed as a direct target of caudal-related homeobox 1 (CDX1) transcription factor . SALL4 is aberrantly expressed in the CDX1-positive intestinal metaplasia of the stomach in both humans and mice. CDX1-induced SALL4 converts gastric epithelial cells into tissue stem-like progenitor cells, which then transdifferentiate into intestinal epithelial cells, suggesting that SALL4 is a critical component of CDX1-directed transcriptional circuitry that promotes intestinal metaplasia. Furthermore, SALL4 is found to be regulated by microRNA in glioma cells . MiR-107 mimics reduce while miR-107 inhibitors increase the SALL4 mRNA level. MiR-107 overexpression inhibits cell proliferation and induces apoptosis in glioma cells, which are reversed by SALL4 reintroduction. An obvious inverse correlation between miR-107 expression and SALL4 level is observed in clinical glioma samples, indicating that upregulation of miR-107 inhibits glioma cell growth through direct targeting of SALL4.
SALL4B can be modiﬁed by both ubiquitination and sumoylation at the post-translational level. However, SALL4B sumoylation is independent of ubiquitination while lysine residues 156, 316, 374, and 401 are essential for sumoylation. It is known so far that only SALL4 sumoylation is functionally important . A constitutive sumoylation of SALL4B is readily detectable in teratocarcinoma cells. SUMO-deﬁciency compromises the transactivational or trans-repressional activities of SALL4B, suggesting that sumoylation is an important post-translational modiﬁcation for SALL4 activity.
SALL4 as cancer biomarker and target
SALL4 seems to be a sensitive and speciﬁc cancer biomarker. SALL4 expression is reported in numerous malignancies, such as precursor B-cell lymphoblastic lymphoma [50,51], myelodysplastic syndromes , acute myeloid leukemia , chronic myeloid leukemia , breast cancer , lung cancer [31,53], endometrial cancer , liver cancer [33,38], gastrointestinal carcinoma [34,54–56], glioma [48,57], germ cell tumors (GCTs), and yolk sac tumors [58–60]. SALL4 expression correlates with disease progression in human AML and its expression in AML patients is correlated with treatment status. Therefore, SALL4 may be used as a marker for diagnosis and prognosis for AML. SALL4 is expressed at a high level in the early clinical stages of breast cancer, indicating that SALL4 may be helpful for breast cancer screening.
SALL4 expression is reactivated in human HCC patients. HCC patients with high SALL4 expression are signiﬁcantly associated with shorter survival. SALL4 is an independent prognostic factor for overall survival and early recurrence of HCC. In endometrial cancer patients, the level of SALL4 expression is positively correlated with worse patient survival and aggressive features such as metastasis. In colorectal carcinoma (CRC), signiﬁcant increase in SALL4 expression is detected in 87 tissues and SALL4 expression is highly correlated with tumor metastasis. In esophageal carcinoma (ESCC), SALL4 expression has a signiﬁcant correlation with invasion and metastasis of the disease . We have shown that SALL4 expression is up-regulated in gastric cancer patients and high level of SALL4 predicts poor prognosis in these patients. Furthermore, a high level of SALL4 protein is detected in the serum of HCC patients . HCC patients with high SALL4 serum levels have poor prognosis evidenced by both tumor recurrence and overall survival rate, suggesting that the high serum level of SALL4 is a novel prognosis biomarker for HCC patients.
In addition to being a biomarker for cancer diagnostics, SALL4 may also constitute a possible therapeutic target. The inhibition of SALL4 expression by siRNA causes reduced cell survival and impaired migration and invasion in distinct cancer cells in vitro.Stable knockdown of SALL4 by shRNA eﬃciently retards tumor growth and restrains tumor metastasis in animal models. Moreover, SALL4 silencing by miRNA inhibits glioma cell proliferation and induces apoptosis in vitro and in vivo. These ﬁndings suggest that depletion of SALL4 has potential anti-tumor effects. In addition, natural compounds such as matrine and apigenin have been shown to suppress SALL4 expression and inactivate the β-catenin signaling pathway in leukemic cells . However, the speciﬁcity of these natural compounds to SALL4 still needs to be further investigated.
SALL4 also serves as an ideal target for combined therapy. SALL4 knockdown in combination with Bcl2 inhibitor treatment increases the apoptotic AML cells to 2–3 fold compared to cells treated alone , suggesting that the combination of Bcl2 inhibitor and down-regulation of SALL4 could be a novel therapeutic strategy in treating AML patients. In human hepatocellular carcinoma, HDAC inhibitor SBHA reduces SALL4 expression and inhibits the proliferation of SALL4-positive HCC cells, suggesting the therapeutic potential of these inhibitors in the treatment of SALL4-positive HpSC HCC through targeting SALL4 . Furthermore, interfering with the interaction of SALL4 and its epigenetic partner complex has therapeutic effects in cancer.
Gao and colleagues have developed a peptide inhibitor that can compete with SALL4 in interacting with the HDAC complex . They demonstrate that treating SALL4 expressing leukemic cells with this peptide leads to cell death through the reactivation of PTEN. The antileukemic effect of this peptide can be conﬁrmed on primary human leukemia cells in culture and in vivo, and is identical to that of down-regulation of SALL4 in these cells by using siRNA. The biological activity of this peptide is further conﬁrmed in hepatocellular carcinoma.TheSALL4 peptide inhibitor inhibits the viability of SALL4-overexpressing hepatocellular carcinoma cells in a PTEN-dependent manner, with minimaltoxiceffectsonSALL4-negativecells,as compared with the HDAC inhibitor trichostatin (TSA). To test the therapeutic effect of this peptide for in vivo treatment, the peptide is conjugated with the transactivator of transcription (TAT) protein transduction domain and intraperitoneally injected into NOD/SCID mice bearing subcutaneous xenograft tumor . TAT fusion peptide greatly reduces the tumorigenicity of SALL4-overexpressing hepatocellular carcinoma cells, suggesting that the SALL4 peptide inhibitor is a potent anti-cancer agent for SALL4-overexpressing hepatocellular carcinoma.
Conclusions and future directions
Over the past decade, accumulating evidence indicates that SALL4 plays a key role in cancer biology in addition to its seminal function in stem cells and development. In distinct types of cancer,SALL4 has been shown to be crucial for cell survival and proliferation, invasive migration, and chemoresistance. Evidence from mouse models suggests that SALL4 is critically involved in leukemogenesis. SALL4 protein may either function as a transcription activator or suppressor to regulate the expression of downstream target genes. However, the pathological roles of SALL4 in cancer seem to be dependent on cell type and context. Therefore, it is necessary to establish mouse models in which SALL4 isoforms are conditionally activated or knocked down in certain cell types. In addition, the oncogenic functions of SALL4 have not been completely characterized. Much less is known about the roles of SALL4 in the other hallmarks of cancer such as sustained angiogenesis, immune evasion, and deregulated energy metabolism. The underlying molecular mechanisms responsible for the different functions of SALL4 in tumor development and progression have not been fully elucidated. Up to now only a few mediators have been identiﬁed.
It is conceivable that many other downstream targets of the SALL4 signaling pathway remain to be discovered. Thus, transcriptomic and proteomic analyses to reveal the global downstream target genes (including both coding and noncoding genes) and interacting proteins for SALL4 will help establish the integrated signaling network in cancer. Moreover, the mechanisms driving the re-activation of SALL4 in cancer remain largely unknown. The activation of SALL4 upstream regulators is frequently seen in human cancers. For instance, STAT3 is associated with constitutive SALL4 expression in breast cancer and inhibition of STAT3 activity disrupts SALL4 expression . Thus, we need to understand if any other genetic or epigenetic modiﬁcations of SALL4 gene exist that contribute to tumor development and progression. In addition, a speciﬁc peptide inhibitor for SALL4 has shown promising anti-cancer activity and further efforts to develop small molecule peptide mimetics or combine with conventional anticancer drugs may help rediscover its therapeutic value. Targeted delivery of siRNA and other inhibitors to disrupt the expression and function of SALL4 in cancer cells by using advanced biotechnologies will provide a new strategy for cancer therapy. Finally, SALL4 is known to maintain the self-renewal and pluripotency of stem cells.
It is interesting to test whether targeting SALL4 will be able to eradicate CSCs given that CSCs are thought to cause metastasis, chemoresistance, and subsequently tumor recurrence. Answers to these important questions will shed light on the role of SALL4 in cancer biology and provide full potential for SALL4 as a valid cancer biomarker and target.
7.11.6 Sal-like 4 (SALL4) suppresses CDH1 expression and maintains cell dispersion in basal-like breast cancer
- SALL4 suppresses CDH1 transcription.
• SALL4 positively regulates expression of ZEB1, a CDH1 suppressor.
• SALL4 prevents intercellular adhesion in basal-like breast cancer.
• SALL4 maintains cell motility in basal-like breast cancer.
In cell cultures, the dispersed phenotype is indicative of the migratory ability. Here we characterized Sal-like 4 (SALL4) as a dispersion factor in basal-like breast cancer. Our shRNA-mediated SALL4 knockdown system and SALL4 overexpression system revealed that SALL4 suppresses the expression of adhesion gene CDH1, and positively regulates the CDH1 suppressor ZEB1. Cell behavior analyses showed that SALL4 suppresses intercellular adhesion and maintains cell motility after cell–cell interaction and cell division, which results in the dispersed phenotype. Our findings indicate that SALL4 functions to suppress CDH1 expression and to maintain cell dispersion in basal-like breast cancer.
Cell migration is recognized in various fields, including cancer. A hallmark of migratory cells is the dispersed phenotype in in vitro condition, in which a cell located at the edge of a cluster loses intercellular adhesion, possesses membrane spikes and front–rear polarity, and moves away independently from the cluster. In contrast, plated non-migratory cells form compacted clusters, where adhesiveness is augmented, and single dispersed cell is not seen. One of the phenomena to induce the migratory ability is epithelial mesenchymal transition (EMT), by which epithelial properties, e.g., the compacted morphology and epithelial marker expression, are replaced by the dispersed phenotype and mesenchymal gene expression . An advantageous model to study cell dispersion and EMT is basal-like breast cancer. Some of basal-like breast cancer cell lines, such as SUM159 and MDA-MB-231, have the dispersed phenotype and mesenchymal gene expression. These characteristics are convertible to epithelial properties by genetic manipulation, which allows us to digest what factor(s) functions to control cell dispersion and EMT. For instance, the zinc finger- and homeobox containing transcription factor ZEB1 (also known as deltaEF1 and TCF8) acts as an EMT activator. ZEB1 suppresses the transcription of the adhesion gene CDH1, and ZEB1 knockdown enhances cell–cell adhesion . The miR200 family of microRNAs is known as a suppressor of the ZEB family . Introduction of miR200-mediated ZEB1 silencing diminishes the dispersed phenotype and motility in MDA-MB-231 . CDH1 encodes the cell–cell adhesion protein E-cadherin. MDA-MB-231 having ectopic E-cadherin expression exhibits the compacted epithelial morphology and loss of the migratory ability . These revealed that CDH1 suppression by ZEB1 plays a key role in the maintenance of cell dispersion in basal-like breast cancer.
Sal-like 4 (SALL4) is one of the mammalian homologs of the Drosophila region specific homeotic genespalt (sal), which encodes a multiple zinc finger transcription factor. SALL4 consists of four exons, and the second of which has an internal splicing donor site. SALL4A, one of two SALL4 variants, is translated from the mRNA having the entire exon2, whereas the mRNA for SALL4B has the short form of exon2 . SALL4 has been identified as a causative factor in acute myeloid leukemia . An increase in SALL4 expression has also been reported in breast-  and , lung- , colorectal-  and liver cancers  as well as germ cell tumors  and . In addition to in cancerous tissues and cancer cell lines, SALL4 expression has been detected in circulating breast cancer cells . In breast cancer cell lines, SALL4 transcription is positively regulated by STAT3 , and SALL4 suppression provides proliferative inhibition  and .
In this study, we identified SALL4 as a cell dispersion factor. We demonstrated that basal-like breast cancer cell lines undergo transition to a compacted epithelial state by SALL4 knockdown. In reciprocal experiments, the overexpression of SALL4 provided the dispersed phenotype and a reduction in CDH1expression to epithelial cells. The time-course observation revealed that SALL4 prevents cell–cell adhesion, and maintains cell motility in basal-like breast cancer.
Epithelial transition is induced by SALL4 knockdown in basal-like breast cancer
SALL4 involves in cell proliferation in breast cancer cell lines  and . The functions of SALL4, however, remain elusive. To analyze the functions of SALL4 in breast cancer, we established a DOX inducible shRNA expression system with shGFP and shSALL4 constructs in the basal-like breast cancer cell line SUM159. To evaluate effects of our system, we analyzed the cell proliferative ability, a known function of SALL4. In our system, reduced cell number was observed in cells having shSALL4#3 and #5 expression, but not in shGFP expression (Fig. S1A–C). Target sites of shSALL4#3 and #5 were designed at the regions common to the mRNAs of SALL4 variants. Because the shSALL4#5 is more effective than the #3, we mainly used the #5 in this study. Quantitative RT-PCR and immunoblotting showed significant reductions in SALL4 mRNA and protein levels in DOX-induced cells (Fig. S1D and E). The ratio between the numbers of dead cells and total cells in shSALL4-expressing cells was identical to that in the no-DOX control (Fig. S1F), indicating that the reduced cell number observed by SALL4 knockdown is not due to decreased cell survival. To analyze changes in expression of the proliferation genes, we quantified the mRNA levels.BMI1, a polycomb group gene, is positively regulated by SALL4 . BMI1 suppresses expression of the cyclin-dependent kinase inhibitors, such as p16, p18 and p21 . In our system, shSALL4 reduced theBMI1 level and increased the p16 and p18 levels ( Fig. S2A–D). We analyzed other proliferation markers,MYC, CCNE1 and CCND1. It has been reported that SALL4 positively regulates CCND1 in transcription level . We observed a reduction in expression of CCND1 in shSALL4-expressing cells ( Fig. S2E–G). In protein analysis, Cyclin D1, the product of CCND1, level was reduced ( Fig. S2H). These results indicate that SALL4 regulates cell proliferation in breast cancer, and our inducible shRNA expression system is useful to explore SALL4 functions.
In in vitro conditions, some of basal-like breast cancer cell lines, including SUM159, tend to be dispersed. Surprisingly, almost cells having shSALL4 expression lost membrane spikes, and formed compacted clusters (Fig. 1A–F). In order to examine this difference, we measured the lengths of perimeters and contacting areas of cells located at the edges of the clusters (Fig. S3). Polarized and spine-rich cells typically have a longer perimeter than spineless cells. We compared the lengths of perimeters of shSALL4-expressing cells to that of no-DOX and shGFP controls. Small values observed in shSALL4-expressing cells indicate that the cells became spineless (Fig. 1G). The ratio of the length of contacting area to that of perimeter reflects the degree of compaction. Cells having shSALL4 expression were more compacted than the controls (Fig. 1H).
Fig. 1. The compacted phenotype and epithelial gene expression observed by SALL4 knockdown in SUM159. (A–F) Cell compaction was observed in cells having shSALL4 expression.
Since mammary cells possess a potential to shift between compacted epithelial and dispersed mesenchymal states  and , the compaction observed in shSALL4-expressing cells was suggestive of a transition to the epithelial state. Thus, we analyzed mRNA levels of the epithelial marker CDH1 and the mesenchymal markers VIM and CDH2 ( Fig. 1I–K). In shSALL4-expressing cells, the CDH1 level was increased and the VIM level was reduced. The CDH2 level was not significantly changed. We detected immunoreaction of E-cadherin, the product of CDH1, in shSALL4-expressing cells ( Fig. 1L–O). Our observations, the compacted phenotype ( Fig. 1D, F and H) and the up-regulation of epithelial markerCDH1 ( Fig. 1I and O), indicate that SALL4 knockdown induces the epithelial transition. The previous study has demonstrated that ectopic E-cadherin expression induces the compacted phenotype and reduction in the vimentin, the product of VIM, level in basal-like breast cancer MDA-MB-231 . SALL4 might regulate cell dispersion and mesenchymal gene expression by suppressing CDH1 transcription.
SALL4 regulates the EMT factor ZEB1
We suspected that SALL4 regulates transcription factors involving in EMT, because SALL4 knockdown induces epithelial transition. In order to identify the factor(s), we used quantitative RT-PCR to screen the transcription factors, SNAI1, SNAI2, TWIST1, TWIST2, FOXC1, FOXC2, TGFB1, TCF3, GSC, GRHL2,ZEB1 and ZEB2 ,  and . In the result, we found reduction in the ZEB1 mRNA level in shSALL4-expressing cells ( Fig. 2A), while the others were not significantly changed ( Fig. 2B and Fig. S4). No detectable amplification was observed in the experiments for GSC and GRHL2. In addition to change in theZEB1 mRNA level, ZEB1 protein level was reduced in shSALL4-expressing cells ( Fig. S5).
Fig. 2. The SALL4-ZEB1 network in SUM159.
ZEB1 mRNA is one of known targets of the miR200 family-mediated gene silencing  and . Thus we assessed the activities of miR200s by using a miR200 reporter ( Fig. S6), which has the ZEB1 3′ untranslated region, a target of miR200 family. To evaluate the miR200 reporter, we introduced expressions of two miR200 regions, miR200b-a-429 and miR200c-141. The expression of miR200 family decreased the luciferase activity ( Fig. 2C), indicating that using the miR200 reporter enables us to examine the activities of miR200s-mediated gene silencing. Comparing to the shGFP control, shSALL4-expressing cells showed no alteration of the luciferase activity ( Fig. 2D). This result indicates that the activities of miR200s are not changed by SALL4 knockdown. It is known that expressions of the ZEB family and the miR200 family are mutually exclusive . ZEB1 and ZEB2 bind to the promoter regions of miR200 family, and suppress their transcription. The miR200s act as the silencer for ZEB1 and ZEB2 mRNAs by binding to their 3′ untranslated regions. A previous study has demonstrated that ZEB1 knockdown increases miR200s activities . We showed reduction in ZEB1 level ( Fig. 2A). However the activities of miR200s were not increased ( Fig. 2D). No alteration of miR200s activity observed in shSALL4-expressing cells is likely due to the function of another miR200s suppressor ZEB2, the mRNA level of which was not changed by SALL4 knockdown ( Fig. 2B). A similar observation has been reported in a study in ovarian cancer, which showed that the miR200c-141 level was not altered by ZEB1 knockdown in cells expressing ZEB2 . To analyze whether ZEB1 promoter activity was affected by SALL4 knockdown, we connected the 5 kbp upstream region of the ZEB1 initiation codon to the luciferase2 gene ( Fig. S6). Cells having thisZEB1 promoter construct showed a reduction in the luciferase activity when shSALL4 was expressed ( Fig. 2E), suggesting that SALL4 positively regulates ZEB1 transcription.
Our results demonstrated that SALL4 regulates two transcriptional regulators, BMI1 ( Fig. S2A) and ZEB1 (Fig. 2A). To analyze whether BMI1 regulates ZEB1 transcription, we performed shRNA-mediated BMI1knockdown assays. Due to severe proliferative inhibition of BMI1 knockdown , we could not obtain enough number of shBMI1 infectants to analyze the gene expression. We therefore established the DOX inducible shBMI1 expression system to obtain a sufficient number of cells with avoiding the proliferative inhibition. The ZEB1 mRNA level was not changed by shBMI1 induction ( Fig. 2F and G). In head and neck squamous cell carcinoma, CDH1 transcription is suppressed by BMI1, and BMI1 knockdown increases the E-cadherin level . In basal-like breast cancer, however, the CDH1 mRNA level was not affected by shBMI1 expression ( Fig. 2H). This suggests that the mechanism of CDH1 regulation is different among cell types.
Besides analyses in shBMI1 expressing cells, we performed ZEB1 knockdown experiments. The BMI1mRNA level was not affected by ZEB1 knockdown ( Fig. 2I and J). The results of ZEB1 and BMI1 knockdown experiments suggest that these transcriptional regulators are independently regulated by SALL4. Since ZEB1 acts as the suppressor for CDH1 transcription , the CDH1 mRNA level was up-regulated by shZEB1 expression ( Fig. 2K). This suggests that the SALL4-ZEB1 network regulates CDH1transcription.
If CDH1 transcription is suppressed by the SALL4-ZEB1 network, a change in CDH1 level should be observed after a reduction in ZEB1 expression when shSALL4 is induced. To analyze the timing of changes in SALL4, ZEB1 and CDH1 expressions, we performed quantitative RT-PCR in the DOX inducible shSALL4 expression system at time points 0.5, 1, 2 and 4 days post DOX administration. A reduction in theSALL4 mRNA level was observed from 0.5 day ( Fig. 3A). The ZEB1 level was significantly changed from 1 day ( Fig. 3B). An increase in the CDH1 level was observed from 2 days, suggesting that up-regulation ofCDH1 transcription occurs between 1 and 2 days ( Fig. 3C). These results support the suggestion thatCDH1 is suppressed by the SALL4-ZEB1 network in basal-like breast cancer.
Sequential gene expression changes after SALL4 knockdown in SUM159.
Fig. 3. Sequential gene expression changes after SALL4 knockdown in SUM159. (A–C) Quantification of the mRNA levels for SALL4 (A,n = 3), ZEB1 (B, n = 3) and CDH1 (C, n = 3) were performed at 0.5, 1, 2 and 4 days. The same values of no-DOX controls as for the analyses shown in Fig. S1 (SALL4), Fig. 2 (ZEB1) and Fig. 1 (CDH1) were used to calculate relative values. Error bars indicate standard deviations. Asterisks indicate statistical significance. Data between the no-DOX control and each time point were analyzed by the Student’s t-test.
SALL4 maintains cell dispersion and regulates gene expression in MDA-MB-231 as well as in SUM159
The previous study has reported that SALL4 knockdown impairs the proliferative ability in another basal-like breast cancer cell line MDA-MB-231 . To assess the generality of our observation in SUM159, we analyzed the changes in the phenotype and gene expression in MDA-MB-231. Cells having shSALL4 expression lost spikes and exhibited an oval-shape (Fig. 4A and B). However, unlike SUM159, the cells were enlarged. The mean perimeter length of enlarged oval-shaped cells was comparable to that of spine-rich controls (Fig. 4C). SALL4 knockdown increased the degree of compaction in MDA-MB-231 (Fig. 4D). In quantitative RT-PCR analyses for the epithelial and mesenchymal genes, CDH1 expression was augmented, and VIM was reduced ( Fig. 4E and F). The CDH2 level was not significantly changed ( Fig. 4G). The levels of transcriptional regulators BMI1 and ZEB1 were reduced by SALL4 knockdown ( Fig. 4H and Fig. S7). These results, except for the effect on cell size, were similar to the observations in SUM159, suggesting that SALL4 maintains cell dispersion and regulates the expressions of epithelial and mesenchymal genes in basal-like breast cancer cell lines.
Mammary epithelial cells exhibit the dispersed phenotype and express the mesenchymal genes by SALL4 overexpression
HMLE is utilized as an epithelial cell model in breast cancer studies . We overexpressed SALL4 variants, SALL4A and SALL4B, in HMLE to analyze whether SALL4 induces cell dispersion in compacted epithelial cells. The EGFP control showed compacted clusters (Fig. 5A). The clusters of SALL4A and SALL4B overexpressing cells had more spaces than that of EGFP control (Fig. 5B and C arrowheads). These spaces were likely to be caused by a loss of adhesiveness. In SALL4 overexpressions, cells exhibited membrane spikes, and the mean lengths of perimeters were increased (Fig. 5D). The degrees of compaction were reduced (Fig. 5E). These results imply that SALL4 forces cell dispersion in HMLE. However, SALL4 itself is insufficient to induce complete cell dispersion as in basal-like breast cancers, suggesting that other supportive factor(s) is required to induce it.
Fig. 5. The dispersed phenotype and mesenchymal gene expression induced by SALL4 overexpression in epithelial cells.
In basal-like breast cancers, SALL4 knockdown increases the expression of adhesion gene CDH1 and reduces the levels of mesenchymal genes VIM and ZEB1 ( Fig. 1, Fig. 2 and Fig. 4). We analyzed the mRNA levels of CDH1, VIM and ZEB1 in HMLE having SALL4 overexpression. SALL4A and SALL4B reduced the CDH1 level ( Fig. 5F), which might involve in a loss of cell–cell adhesion. Conversely, the VIMand ZEB1 expressions were up-regulated ( Fig. 5G and H). Our results suggest that in addition to the maintenance of dispersed phenotype in basal-like breast cancers, SALL4 is capable of inducing cell dispersion with a reduction in CDH1 expression and an increase in the transcription of mesenchymal genes in epithelial cells. Given that SALL4 up-regulates the ZEB1 transcription ( Figs. 2A and E, 4H, 5H), and that ZEB1 suppresses the CDH1 transcription ( Fig. 2K) , the down-regulation of CDH1 was likely to be caused by an ectopic activation of SALL4-ZEB1 network. Since loss of CDH1 function diminishes intercellular adhesiveness in HMLE , CDH1 suppression by the SALL4-ZEB1 network might result in a loss of adhesiveness. We observed identical effects between the SALL4A and SALL4B overexpressions in HMLE, indicating that the regulation of cell dispersion is a fundamental function of SALL4.
SALL4 suppresses cell–cell adhesion to maintain cell dispersion in basal-like breast cancer
Cells having shSALL4 expression exhibited compacted clusters (Fig. 1D and F), suggesting that SALL4 knockdown changes cell behavior. Time-lapse microscopy is utilized to explore cell movements. We performed the time-lapse analyses from 1 to 2 days after starting incubation with DOX in which compaction is initiated in our SALL4 knockdown system. The up-regulation of CDH1 transcription is also initiated between 1 and 2 days post DOX administration ( Fig. 3C). The no-DOX controls repeated contact and dispersion ( Movie S1). For instance, as shown in Fig. 6A top, one cell interacted with another cell at time point 20 min, and the contact was preserved until 120 min. At 140 min, the cells were uncoupled. Subsequently, one uncoupled cell collided with the other cell at 160 min, and dispersed immediately. In comparison to the control, the contacting period of the cell having shSALL4 expression was extended ( Fig. 6A bottom, Movie S2). Cells were interacted at 20 min, and adhered. This contact was persisted for longer than 200 min.
Fig. 6. Attenuation of the dispersion ability in shSALL4-expressing SUM159
For further understanding of the behavior, we compared the frequencies of cells immediately dispersed, dispersed in 1, 3 and 5 h, and adhered longer than 5 h after cell–cell interaction (Fig. 6B). Most of the control cells were dispersed within 5 h after interaction (78.11%). In shSALL4-expressing cells, the frequencies of dispersion within 5 h were decreased (48.12%), and the rate of formation of intercellular adhesion was increased (21.89–51.88%). In addition, we analyzed the frequencies after cell division (Fig. 6C). Similarly to the results of after cell–cell interaction, the frequencies of dispersion within 5 h were reduced (24.57–5.31%). These suggest that SALL4 knockdown impairs the dispersion ability by enhancing intercellular adhesiveness.
We asked whether shSALL4-expressing cells do not disperse from highly compacted clusters, and whether the compacted clusters move around. We performed the wound healing assay in 80–90% confluent cultures with the proliferation inhibitor mitomycin C (Fig. 7). Because the proliferative ability is different between the control and shSALL4-expressing cells (Fig. S1), inhibition of proliferation was demanded to count the exact number of cells moved into the scratched areas, and to analyze the dispersion ability in the wound healing assay. To determine the concentration of mitomycin C, we performed a growth assay, and found that 0.5 μg/ml of it sufficiently inhibited cell proliferation (Fig. S8). In controls, cells were dispersed and filled the scratched areas (Fig. 7E–G). However the number of cells in the scratched areas was significantly reduced in shSALL4-expressing cells (Fig. 7H and I), indicating that shSALL4-expressing cells do not disperse from their cluster, and that the compacted cluster is immobile. We, moreover, analyzed the speeds of movement in the time-lapse movies used for the analyses shown inFig. 6. The mean and maximum speeds were not changed between single cells with and without shSALL4 expression (Fig. 8A and B, single). We also analyzed the motility of cells in contact with other cell(s). Although the no-DOX control had an identical moving speed to single cells, shSALL4-expressing cells showed reduced moving speeds (Fig. 8A and B contacting). The attenuation of the motility observed in contacting shSALL4-expressing cells is consistent with the results of the wound healing assay, and suggests that the trigger to lose cell motility is intercellular adhesion.
Fig. 7. Loss of migratory ability in compacted shSALL4-expressing SUM159.
Fig. 8. Attenuation of the motility in contacting shSALL4-expressing SUM159.
We showed that the ability of cell–cell adhesion after interaction and cell division was enhanced in shSALL4-expressing cells (Fig. 6A–C). In epithelial cells, the adhesion protein E-cadherin localizes in the contacting area to form cell–cell adhesion after interaction . Our immunostaining for E-cadherin showed strong signals in the contacting areas (Fig. 1O), supporting the notion that cell–cell adhesion is enhanced by SALL4 knockdown. As exemplified by the wound healing assay and the analysis of moving speeds (Fig. 7 and Fig. 8), intercellular adhesion observed in shSALL4-expressing cells persists for more than 24 h, and adhered cells loses their motility. Accumulation of the low-motile adhered cells could develop to compacted clusters. Taken together, SALL4 functions to suppress the formation of cell–cell adhesion to preserve cell motility when cells interact, which contributes to the dispersed phenotype.
In this study, we identified SALL4 as the cell dispersion factor. SALL4 suppresses the adhesion geneCDH1, and positively regulates the CDH1 suppressor ZEB1. Consistent with the previous study , basal-like breast cancer having shSALL4-induced CDH1 expression lost the dispersed phenotype. The STAT3 inhibitor impairs the dispersion ability in glioma cells . Given that STAT3 is a positive regulator forSALL4 transcription in breast cancer , our findings are in agreement with the report in glioma cells. Dispersion from an adhesive cluster is one of the characteristics of metastatic cancer . Some of compacted cancer cells acquire the motility and migrate from a cluster to a distant site. Similar events are known in other research fields, such as migratory neural crest cells in development  and cardiomyocyte migration in regeneration . Therefore, this study might not only contribute to therapies for cancer metastasis, but also facilitate understanding of the nature of cell migration.
7.11.7 The transcription factor SALL4 regulates stemness of EpCAM-positive hepatocellular carcinoma
Background & Aims: Recent evidence suggests that hepatocellular carcinoma can be classified into certain molecular subtypes with distinct prognoses based on the stem/maturational status of the tumor. We investigated the transcription program deregulated in hepatocellular carcinomas with stem cell features. Methods: Gene and protein expression profiles were obtained from 238 (analyzed by microarray), 144 (analyzed by immunohistochemistry), and 61 (analyzed by qRT-PCR) hepatocellular carcinoma cases. Activation/suppression of an identified transcription factor was used to evaluate its role in cell lines. The relationship of the transcription factor and prognosis was statistically examined. Results: The transcription factor SALL4, known to regulate stemness in embryonic and hematopoietic stem cells, was found to be activated in a hepatocellular carcinoma subtype with stem cell features. SALL4-positive hepatocellular carcinoma patients were associated with high values of serum alpha fetoprotein, high frequency of hepatitis B virus infection, and poor prognosis after surgery compared with SALL4-negative patients. Activation of SALL4 enhanced spheroid formation and invasion capacities, key characteristics of cancer stem cells, and up-regulated the hepatic stem cell markers KRT19, EPCAM, and CD44 in cell lines. Knockdown of SALL4 resulted in the down-regulation of these stem cell markers, together with attenuation of the invasion capacity. The SALL4 expression status was associated with histone deacetylase activity in cell lines, and the histone deacetylase inhibitor successfully suppressed proliferation of SALL4-positive hepatocellular carcinoma cells. Conclusions: SALL4 is a valuable biomarker and therapeutic target for the diagnosis and treatment of hepatocellular carcinoma with stem cell features.
7.11.8 Overexpression of the novel oncogene SALL4 and activation of the Wnt.β-catenin pathway in myelodysplastic syndromes
Myelodysplastic syndromes (MDS) are a group of heterogeneous clonal stem cell diseases with a tendency to progress to leukemic transformation. The cytogenetic and molecular pathogenesis of MDS has not been well understood. SALL4, a newly identified oncogene, modulates stem cell pluripotency and self-renewal capability in embryonic development and also plays a role in leukemogenesis. Overexpression of SALL4 induces MDS-like features and subsequent leukemic progression in transgenic mice. Here, we examined SALL4 expression levels in bone marrow mononuclear cells from MDS patients, acute myeloid leukemia (AML) patients, and normal control subjects using a semiquantitative reverse transcription polymerase chain reaction. Higher levels of SALL4 expression were seen in MDS and AML samples than in control samples. The expression level of SALL4 positively correlated with those of MYC and CCND1, both of which are downstream target genes in the Wnt/beta-catenin pathway. We therefore propose that SALL4 plays a critical role in the pathogenesis of MDS by causing the aberrant activation of the Wnt/beta-catenin pathway.
Wnt Signaling Inhibition: Will Decades of Effort Be Fruitful at Last?
Emma Hitt, PhD
Oncology Live Published Online: Monday, January 7, 2013
The Wnt signaling pathway was first characterized in the 1970s in Drosophila melanogaster development. It was later recognized in mammalian systems for its importance in cancer. Specifically, core components of this pathway were shown to be dysregulated in colorectal disease. It has since been shown to play a role in other forms of cancer, where it promotes proliferation and survival. It also plays an important role in the maintenance of the pool of tumor-initiating cells, which promote the regrowth of tumors after an insult like surgery or chemotherapy. Tumor-initiating cells are thought to be the source of metastasis. For these reasons, the Wnt pathway is viewed as a strong candidate for therapeutic intervention.
Wnt Signaling in Cancer
The Wnt pathway takes on many forms that fall under the broad classifications of canonical and noncanonical. The canonical pathway deals with the regulation of β-catenin protein levels. Under normal conditions, a cytosolic scaffold known as the destruction complex binds and phosphorylates β-catenin, resulting in its ubiquitylation and degradation. The destruction complex includes the adenomatous polyposis coli (APC) protein, axin, and glycogen synthase kinase-3β. When Wnt ligand binds to the Frizzled receptor, its coreceptor LRP5/6 is recruited and phosphorylated in the intracellular domain, promoting the binding of Dishevelled protein and the sequestration of axin. This disintegrates the destruction complex, resulting in accumulation of β-catenin in the cytosol and upregulated trafficking into the nucleus. β-catenin promotes transcription of genes related to proliferation and survival by acting as a coactivator for the Tcf/Lef family of transcription factors in the nucleus.
Aside from canonical Wnt signaling, two major noncanonical pathways have been studied. In the first of these two pathways, Wnt ligand binding to the Frizzled receptor induces recruitment of Dishevelled protein and the Dishevelled-associated activator of morphogenesis 1 (Daam1). This complex initiates a cascade that activates the Rac and Rho GTPases to mediate cell polarity. The other most widely studied noncanonical Wnt signaling pathway is related to calcium signaling. Wnt ligand binding to the Frizzled receptor promotes the recruitment of Dishevelled in complex with a G protein. This complex promotes intracellular calcium levels to mediate other signaling pathways.
Biological systems tightly regulate the Wnt signaling pathway to prevent aberrant cell growth. It has been known for decades that dysregulation of Wnt signaling leads to cancer, where it was first recognized in familial colorectal disease with mutations in the APC gene. Since then, Wnt signaling has been found to act prominently in breast, liver, skin, and prostate cancers.
Aberrations in canonical Wnt signaling can manifest in many ways. For example, proteins involved in the destruction complex can become nonfunctional through mutations or truncations, inhibiting β-catenin downregulation. The β-catenin protein itself can be mutated to inhibit its recognition by the destruction complex. In addition, the production of Wnt ligand or receptors can be upregulated, resulting in excessive signaling. These different routes of activation complicate the use of a single therapeutic against the Wnt pathway.
Wnt Signaling Pathways
This illustration depicts the best-elucidated cancer-promoting routes of Wnt cell signaling, which draws its name from the wingless mutation in the fruitfly Drosophila melanogaster.
Therapeutic Targeting of Wnt Pathway
Wnt signaling is of great interest to cancer researchers because it is linked to many different forms of the disease. Preclinical models throughout the last decade have established this pathway as an attractive drug target. However, to date, therapies meant to attenuate the Wnt pathway have remained largely theoretical and preclinical. Thankfully, compounds are now starting to enter clinical trials.
Vitamin D has been postulated as a suitable anti-Wnt therapy. The vitamin D receptor binds and sequesters β-catenin at the plasma membrane, inhibiting its nuclear translocation. Mice bearing an APC mutation that promotes spontaneous colon cancer developed more disease when the vitamin D receptor was knocked out. Additionally, the association between sunlight exposure and decreased risk of colon cancer implies that the inhibition of Wnt signaling by vitamin D may be conserved in humans. This phenomenon has yet to be tested in a systematic trial, however.
High-throughput screens have been used extensively to identify small-molecule targeted inhibitors of different Wnt pathway constituents. The binding of β-catenin to its nuclear targets T-cell factor (Tcf) or Creb-binding protein (CBP) is a prototypical example of study in this realm. Screens were performed to identify specific inhibitors of this interaction. This work has identified useful hits in cell-based assays, but translation into the clinic has remained difficult due to the potential off-target effects. First, the drugs identified so far fail to discriminate between the binding of β-catenin to Tcf or APC, so the drug may prevent degradation of β-catenin in addition to its intended effect on transcription. Disruption of the destruction complex binding to β-catenin could lead to severe side effects in normal tissue. Another deleterious effect identified with blockers of β-catenin binding is the inhibition of complex formation with E-cadherin at cell-cell junctions. The net effect of this phenomenon could be to impair cell adhesion on a grand scale. As a result of these challenges, only one compound that directly disrupts β-catenin function has moved into clinical trials.
Other drug candidates inhibit Wnt aberrations upstream of β-catenin function. Several of these compounds have recently begun clinical trials. They block either Wnt ligand secretion or recognition, and preclinical evidence has been encouraging so far. Targeting at this level can also lead to side effects, however. By blocking Wnt signaling at the level of the membrane, it is possible to inhibit noncanonical in addition to canonical signaling. The effect this will have on the body is unknown. In addition, blocking Wnt ligand-receptor interactions may not be sufficient to inhibit Wnt signaling since the activating event may be mutation within the destruction complex or β-catenin itself. In these settings, Wnt signaling can be rendered constitutive and independent of ligand, and the therapy would likely fail.
Moving Into the Clinic
No approved compounds exist for the treatment of Wnt signaling. Phase I trials for these inhibitors should be illustrative in the coming years. While many cancers are addicted to this signaling for long-term growth and renewal, high-turnover tissues like the gastric epithelium and hair follicles are similarly reliant. Therefore, the field is cautious when utilizing any blockade of Wnt signaling, as significant toxicity may result.
If Wnt pathway inhibitors can be proven safe, it may represent a milestone in cancer research given the strong preclinical evidence for cancer cell cytotoxicity. Since this pathway is also crucial in the maintenance of tumor-initiating cells, inhibition could represent a powerful tool in our arsenal to target cells that are resistant to traditional chemotherapy and promote metastasis.
Wnt-Targeting Compounds in Development
|Phase I/II Trials|
(OncoMed Pharmaceuticals/ Bayer)
|This monoclonal antibody targets the Frizzled receptors to block association with Wnt ligands. It was recently shown to potently block the capabilities of pancreatic tumor-initiating cells in a serial dilution assay. In xenograft models of breast, lung, pancreatic, and colon cancer, OMP-18R5 demonstrated significant inhibition of tumor growth, and it synergized with standard-of-care treatment in these models (paclitaxel in breast cancer, for example). (NCT01345201)|
(OncoMed Pharmaceuticals/ Bayer)
|This agent is a fusion protein of the Frizzled8 ligand-binding domain with the Fc region of a human immunoglobulin. It binds and sequesters soluble Wnt ligand, impairing its recognition by receptors on tissues. (NCT01608867)|
(Prism Pharma Co, Ltd/Eisai)
|This is a small-molecule inhibitor of the interaction between β-catenin and CBP. Disrupting the interaction prevents activated transcription by aberrant Wnt signaling at many levels. It is being studied in both solid tumors and myeloid malignancies. (NCT01606579, NCT01302405)|
|This small molecule inhibits acyltransferase porcupine. Preclinical work demonstrated this enzyme’s action is crucial in the secretion of Wnt ligands out of the cell; therefore, inhibiting porcupine can be a small-molecule–based method for inhibiting Wnt ligand-mediated activation. (NCT01351103)|
|This small-molecule poly(ADP ribose) polymerase (PARP) inhibitor has demonstrated efficacy in cellular models of cancer survival. In Wnt signaling, PARPs like the tankyrases promote the ribosylation and subsequent degradation of axin, a key scaffolding protein of the destruction complex. By inhibiting tankyrase, the axin protein is stabilized and can promote the degradation of β-catenin.|
|This selective tankyrase1/2 inhibitor has been shown to inhibit the growth of colon cancer cells in both cell and animal models.|
Sources: ClinicalTrials.gov website, company websites.
- Chen B, Dodge ME, Tang W, et al. Small molecule-mediated disruption of Wnt-dependent signaling in tissue regeneration and cancer [published online ahead of print January 4, 2009]. Nat Chem Biol. 2009;5(2):100–107. doi: 10.1038/nchembio.137.
- Clevers H, Nusse R. Wnt/β-catenin signaling and disease. Cell. 2012;149(6):1192-1205.
- Eckhardt SG. Targeting the WNT Pathway for Cancer Therapy. Presented at: 10th International Congress on Targeted Therapies in Cancer; August 17-18, 2012; Washington, DC.
- He B, Reguart N, You L, et al. Blockade of Wnt-1 signaling induces apoptosis in human colorectal cancer cells containing downstream mutations. Oncogene. 2005;24(18):3054-3058.
- Ichii S, Horii A, Nakatsuru S, et al. Inactivation of both APC alleles in an early stage of colon adenomas in a patient with familial adenomatous polyposis (FAP). Hum Mol Genet. 1992;1(6):387-390.
- Korinek V, Barker N, Morin PJ, et al. Constitutive transcriptional activation by a beta-catenin-Tcf complex in APC-/- colon carcinoma. Science. 1997;275(5307):1784-1787.
- Rubinfeld B, Souza B, Albert I, et al. Association of the APC gene product with beta-catenin. Science. 1993;262(5140):1731-1734.