Oncogene (2012) 31, 4233–4244 & 2012 Macmillan Publishers Limited All rights reserved 0950-9232/12 www.nature.com/onc ORIGINAL ARTICLE Dickkopf 4 (DKK4) acts on Wnt/b-catenin pathway by influencing b-catenin in hepatocellular carcinoma

S Fatima1,7, NP Lee1,7, FH Tsang1, FT Kolligs2, IOL Ng3, RTP Poon1, ST Fan1 and JM Luk1,4,5,6

1Department of Surgery, The University of Hong Kong, Pokfulam, Hong Kong; 2Department of Medicine II, Klinikum Grosshadern, University of Munich, Marchioninistrasse, Munich, Germany; 3Department of Pathology, The University of Hong Kong, Pokfulam, Hong Kong; 4Department of Pharmacology and Department of Surgery, National University of Singapore, Singapore; 5Cancer Science Institute and Institute of Molecular and Cell Biology, Singapore and 6Department of Oncology, pRED China, Roche R&D Center, Shanghai, China

Deregulation of Wnt/b-catenin pathway is a hallmark of Introduction major gastrointestinal cancers including hepatocellular carcinoma (HCC). The oncogenic role of b-catenin is well Liver cancer ranks as the second and sixth leading cause defined but reasons for its accumulation in HCC remain of cancer death worldwide in men and women, unclear. Dickkopf 4 (DKK4) acts as a negative regulator respectively, and hepatocellular carcinoma (HCC) of Wnt/b-catenin pathway but its functional role in liver accounts for 70–80% of total liver cancers (Jemal carcinogenesis has not been studied. We investigated the et al., 2011). HCC associates a dismal prognosis because role of DKK4 in b-catenin regulation in HCC. Reduced of late presentation of the disease and high risk of expression of DKK4 was found in 47% (38/81) of HCC, postoperative tumour recurrence (Hao et al., 2009). as measured by quantitative real time PCR. Ectopic Although the risk factors for HCC are well known expression of DKK4 in two HCC cell lines, PLC/PRF/5 (Kemp et al., 2005), the cellular and molecular mechan- (PLC) and MHCC97L (97L), attenuated b-catenin isms contributing to hepatocarcinogenesis are poorly responsive luciferase activity, and decreased both understood. The Wnt/b-catenin pathway is critical for b-catenin and cyclin D1 protein levels. To study the effect embryonic development, cell proliferation, survival, self- of DKK4 on cell growth and tumourigenicity, two stable renewal and regeneration (Gonzalez, 2006; Nejak-Bowen HCC cell lines were established from PLC and 97L cells. and Monga, 2011). However, its deregulation is a Functional assays demonstrated that overexpression of hallmark of several cancers including HCC (Inagawa DKK4 hampered cell proliferation, reduced colony et al., 2002). Surplus b-catenin in the cytoplasm is formation and retarded cell migration. When DKK4- phosphorylated by a multiprotein destruction complex expressing 97L stable cells were used to induce tumour consisting of glycogen synthase kinase 3b (GSK3b), Axin xenografts in nude mice (n ¼ 8), reduction in tumour sizes and adenomatous polyposis coli (APC). The phosphory- was observed (P ¼ 0.027). Furthermore, immunohisto- lated b-catenin is then ubiquitinated and degraded by the chemical studies showed that decreased expression of proteasome (Orford et al., 1997; MacDonald et al., DKK4 was associated with b-catenin accumulation in 2009). Upon binding of Wnt to one of the members of HCC tissues. Additionally, inhibition of the proteasome the frizzled family receptors and to the low-density using specific inhibitor in DKK4-expressing 97L stable lipoprotein-related protein-5 (LRP5) or LRP6 co-recep- cells masked the effect of b-catenin. Our findings suggest tors, the phosphorylation and degradation of b-catenin is a potential tumour suppressive role of DKK4 as well as inhibited, and it accumulates in the cytoplasm and that of an important regulator of HCC. translocates to the nucleus (Zeng et al., 2005). Inside the Oncogene (2012) 31, 4233–4244; doi:10.1038/onc.2011.580; nucleus, b-catenin forms a complex with T-cell factor published online 16 January 2012 (TCF)/lymphoid enhancer factor and upregulates target genes that are implicated in cancer development for Keywords: hepatocellular carcinoma; Wnt/b-catenin example, c-Myc and cyclin D1 (Logan and Nusse, 2004). signalling; DKK4; tumourigenesis Although Wnt/b-catenin signalling is frequently activated in HCC, reasons for its activation are not well understood. Somatic mutations of b-catenin have only been observed in 12–26% of HCC (de La Coste et al., 1998; Wong et al., 2001), but cytoplasmic and nuclear accumulation of b-catenin have been reported Correspondence. Current address: Professor JM Luk, Department of more commonly in 40–70% of HCC (Huang et al., 1999; Oncology, Roche R&D Center, pRED China, 720 Cai Lun Road, Wong et al., 2001). Mutations of other Wnt/b-catenin Shanghai 201203, China. signalling components like Axin1 and Axin2 have only E-mail: [email protected] been observed in 10% and 3% of HCC cases, 7These authors contributed equally to this work. Received 18 July 2011; revised 17 October 2011; accepted 9 November respectively (Taniguchi et al., 2002). Furthermore, 2011; published online 16 January 2012 inactivating mutations of APC are unlikely to contribute Action of DKK4 on Wnt/b-catenin pathway S Fatima et al 4234 significantly as they are very rare (Chen et al., 1998), frequent in HCC tissues compared with their matched implying that other undefined mechanisms may con- non-cancerous cirrhotic tissues, implying that methyla- tribute to Wnt/b-catenin activation in HCC. tion of DKK2 and DKK3 may not be an early event in The dickkopf (DKK) family is important for verte- HCC but may function in the progression of HCC brate development and comprises four secreted glyco- (Yang et al., 2010). Additionally, methylation of DKK3 proteins (DKK1, DKK2, DKK3 and DKK4) of has been significantly associated with short progression- 255–350 amino acids (Niehrs, 2006). They are classified free survival in HCC patients (Ding et al., 2009; Yang as inhibitors of Wnt/b-catenin signalling but members of et al., 2010). Although reduced expression of DKK4 has the DKK family exert different effects on Wnt/b-catenin only been reported in HCC cell lines (Maehata et al., signalling, and their mechanism of action has not been 2008), its function in hepatocarcinogenesis with respect fully delineated (Fatima et al., 2011). With the exception to the Wnt/b-catenin pathway has not been explored. of DKK3, whose receptor has not been identified, Therefore, in this study we investigated the functional DKK1, DKK2 and DKK4 compete with frizzled family role of DKK4 in HCC. receptors binding to the extracellular domain of LRP5/ LRP6 receptor complex in response to attenuation of Wnt activity (Krupnik et al., 1999; Semenov et al., 2001; Results Niehrs, 2006). DKK1, DKK2 and DKK4 also bind to Kremens, another class of transmembrane receptors Reduced expression of DKK4 in HCC cell lines (Mao et al., 2002; Mao and Niehrs, 2003; Nakamura and tissues and Matsumoto, 2008), although the exact mechanism Eight HCC cell lines were evaluated for DKK4 mRNA of how Krm-DKKs-LRP6 complex inhibit Wnt activity expression using quantitative PCR (qPCR) and com- has not been defined. DKK1 is an antagonist of the pared with immortalized normal hepatocytes MIHA. Wnt/b-catenin pathway and also a downstream target of Marked reduction of DKK4 was found in all HCC cell Wnt/b-catenin signalling (Niida et al., 2004; Chamorro lines (Figure 1a). To test whether the reduced expression et al., 2005; Gonzalez-Sancho et al., 2005). However, in of DKK4 in HCC cells is due to DNA methylation, HCC its negative feedback mechanism is abrogated and representative HCC cell lines, such as Hep3B, HuH-7, overexpression of DKK1 in HCC has been associated PLC/PRF/5 (PLC) and MHCC97L (97L), were treated with cytoplasmic and nuclear b-catenin accumulation with increasing concentrations of 5-azacytidine, a (Yu et al., 2009). DKK2 can serve as both an inhibitor pharmacological inhibitor of DNA methyltransferase and activator depending on cellular context (Mao et al., to rule out the effects of DNA methylation. Figure 1b 2001), and the role of DKK3 remains unclear. Previous illustrates unaltered DKK4 expression following treat- study suggested DKK3 to have no effect on Wnt/b- ment with 5-azacytidine, suggesting other reasons for its catenin signalling (Krupnik et al., 1999), whereas others reduced expression in HCC. We next checked the have demonstrated nuclear accumulation of b-catenin as expression of DKK4 in 81 matched pairs of HCC a result of reduced DKK3 expression in lung cancer tumours and their corresponding adjacent non-tumour (Yue et al., 2008). Furthermore, DKK2 and DKK3 are liver tissues of the same patient. DKK4 was reduced in both methylated in HCC and their methylation is more 47% (38/81) of HCC tumour tissues (Figure 1c).

Figure 1 Reduced expression of DKK4 mRNA in HCC cell lines and tissues. qPCR was used to determine the mRNA expression of DKK4 in HCC cells and tissues. (a) Reduced expression of DKK4 mRNA was found in HCC cells (HepG2, Hep3B, HuH-7, PLC, H2P, H2M, 97L and 97H) when compared with immortalized normal hepatocytes MIHA. (b) This reduction observed in HCC cells was not associated with DNA methylation when representative HCC cells (Hep3B, HuH-7, PLC and 97L) were treated with increasing concentrations of 5-azacytidine and no effects on DKK4 mRNA expression were observed. (c) Reduced mRNA expression of DKK4 was found in tumour (T) tissues when compared with their adjacent non-tumour liver (NT) tissues in 81 patients with primary HCC. DKK4 expression of each sample is presented by the fold ratio against all NT tissues. *Po 0.05.

Oncogene Action of DKK4 on Wnt/b-catenin pathway S Fatima et al 4235 Whereas DKK4-negative was defined as having a DKK4 suppresses in vivo tumourigenicity tumour/non-tumour expression ratio less than twofold, To further explore the in vivo tumour suppressive ability DKK4-positive had this ratio greater than twofold. The of DKK4, tumour formation in nude mice was assessed non-tumour is liver tissue adjacent (at least 2 cm in by s.c. injection of 97L_DKK4 cells and 97L_vec cells. distance) to the HCC tumour, which may be cirrhotic, All mice were observed for 8 weeks. Of considerable chronic hepatitis or apparent normal hepatocytes in interest, tumours derived from 97L_DKK4 cells grew histology. However, we did not find any association substantially slower compared with tumours derived between DKK4 expression and any clinicopathological from 97L_vec cells during the entire tumour growth parameters studied (Supplementary Table 1). period (Figure 4a). Within 10 days, solid tumours were visible in all mice injected with 97L_vec cells. On the contrary, no tumours were observed in mice injected DKK4 attenuates TCF-mediated luciferase activity with 97L_DKK4 cells at 4–5 weeks, and one mouse did and reduces b-catenin protein levels not develop any tumour at 8 weeks, after s.c. injection. To determine whether DKK4 affects b-catenin in HCC, we As shown by qPCR, tumours derived from 97L_DKK4 used a dual luciferase reporter assay with b-catenin had reduced b-catenin and cyclin D1 mRNA levels responsive luciferase vectors, TOPflash and FOPflash. compared with tumours derived from 97L_vec cells PLC and 97L cells (both with high endogenous b-catenin (Figure 4b). These results suggest that DKK4 has an levels) were transfected with plasmid encoding DKK4, and important role in suppressing tumour growth with a the resulting luciferase activities were measured. Our results regulatory effect on b-catenin signalling. Subsequently, revealed that exogenous DKK4 inhibited the transactiva- we performed immunohistochemical analysis for DKK4 tion of TCF reporter mediated by the endogenous and in tumour xenografts and confirmed that DKK4 protein ectopic wild-type b-catenin, but not that of mutant b- was highly expressed in tumours of mice that had catenin (Figure 2a), suggesting that site of b-catenin 97L_DKK4 cells. We next investigated the effect of phosphorylation may be important for DKK4-mediated DKK4 overexpression on the Wnt/b-catenin pathway inhibition. Furthermore, transient overexpression of DKK4 in vivo. Immunohistochemical analysis on tumour reduced b-catenin and cyclin D1 protein levels in PLC and xenografts showed that tumours from 97L_vec cells 97L cells (Figure 2b). The DKK4 antibody specificity was had strong cytoplasmic and nuclear b-catenin accumu- confirmed by peptide blocking experiments as shown by lation as well as strong cyclin D1 expression, whereas western blot and immunohistochemical analysis (Supple- tumours from 97L_DKK4 cells had a reduced b-catenin mentary Figure 1). We next studied the effect of DKK4 on accumulation in both the cytoplasm and the nucleus, cytoplasmic and nuclear levels of b-catenin. Ectopic and these tumours also exhibited reduced cyclin D1 expression of DKK4 in PLC and 97L cell lines resulted in levels (Figure 4c). These results suggest that DKK4 a significant reduction of cytoplasmic b-catenin but the attenuates the Wnt/b-catenin pathway and has an change in nuclear b-catenin protein levels was not important regulatory role in tumourigenesis in HCC. significant as shown in the western blots (Figure 2c). Altogether, these results suggest that DKK4 reduces HCC tissues with b-catenin accumulation show reduced TCF-mediated luciferase activity of wild-type b-catenin DKK4 expression but not that of mutant b-catenin. Furthermore, the The reduced accumulation of b-catenin in tumours of reduction of b-catenin protein levels implicates that mice that had 97L_DKK4 cells, led us to investigate DKK4 may effect as a potential negative regulator to whether DKK4 protein expression affected b-catenin attenuate the Wnt/b-catenin pathway. accumulation in HCC tissues. By immunohistochemical analysis, DKK4 levels were reduced in 24 out of 41 (58%) tumour tissues compared with corresponding DKK4 exhibits tumour suppressive ability adjacent non-tumour liver tissues. b-Catenin accumula- The reduced expression of DKK4 in HCC tissues tion was detected in 29 out of 41 (71%) HCC tumour prompted us to investigate whether DKK4 acts as a tissues compared with corresponding adjacent non- tumour suppressor. Two DKK4-overexpressing stable tumour liver tissues. Nuclear staining of b-catenin cell lines (PLC_DKK4 and 97L_DKK4) and two was detected in 7 cases, with the rest showing strong control-vector stable cell lines (PLC_vec and 97L_vec) staining in the cytoplasm and the cell membrane. were established and DKK4 expression was confirmed Furthermore, of the 24 cases that showed reduced using western blot (Figure 3a). The tumour suppressive DKK4 levels, b-catenin accumulation was detected in 20 function of DKK4 was assessed in these stable cell lines cases (Figure 5) and this was statistically significant by in vitro MTT proliferation assay, colony formation (P ¼ 0.035) (Supplementary Figure 2). assay and wound-healing assay. PLC_DKK4 and 97L_DKK4 cells significantly reduced cell proliferation GSK3b and the proteasome are involved in (Po0.05, Figure 3b) and formed fewer colonies DKK4-mediated b-catenin degradation (Po0.05, Figure 3c) compared with the vector-carrying As DKK4-overexpressing cells reduced b-catenin levels, stable transfectants. The wound-healing assay demon- we further investigated whether DKK4 affected other strated that DKK4-overexpressing stable clones showed components of the Wnt/b-catenin pathway. As shown in retarded migration towards the wound when compared Figure 6a, compared with 97L_vec cells, 97L_DKK4 with the vector control cells (Figure 3d). cells have reduced levels of active form of b-catenin

Oncogene Action of DKK4 on Wnt/b-catenin pathway S Fatima et al 4236 (dephosphorylated on Ser37 or Thr41), which mediates for degradation. Furthermore, the effect of DKK4 on target gene activation. DKK4 overexpression also b-catenin was also blocked in the presence of protea- resulted in an increase in phosphorylated b-catenin some inhibitor MG132, implying that DKK4 promotes (phospho-b-catenin). This form of b-catenin is targeted degradation of b-catenin through the proteasome

Figure 2 DKK4 inhibits TCF reporter activity, and reduces b-catenin and cyclin D1 protein levels. (a) PLC and 97L cells were grown in 6 well-plates and co-transfected with TOPflash or FOPflash, wild-type b-catenin or mutant b-catenin, DKK4 and pRL-TK. Normalized luciferase activity was measured 48 h after transfection, and reduced luciferase activities were observed in cells transfected with DKK4 and wild-type b-catenin, but there was no effect associated with cells transfected with DKK4 and mutant b-catenin. **Po0.005. At 48 h after DKK4 transfection, reduction in b-catenin and cyclin D1 protein levels (b) and cytoplasmic b-catenin levels (c) were observed in PLC and 97L cells using western blot. Western blot bands were quantified by Image J Software National Institutes of Health Bethesda, MD, USA following normalization to b-actin.

Figure 3 Ectopic expression of DKK4 inhibits cell growth, colony formation and cell mobility in cultured HCC cells. (a) Two stable cell lines overexpressing DKK4 (PLC_DKK4 and 97L_DKK4) and control vector (PLC_vec and 97L_vec) were established in PLC and 97L cells. High level of DKK4 protein level was observed in stable cell lines using western blot. Western blot bands were quantified by Image J Software, following normalization to b-actin. (b) Equal numbers of DKK4 and vector-transfected stable cells were plated in 96-well plates, and cell proliferation was quantified by MTT assay for 4 days. Reduced cell proliferation was observed in DKK4-expressing stable cells. *Po0.05, a, P ¼ 0.024, b, P ¼ 0.019, c, P ¼ 0.017, d, P ¼ 0.029, e, P ¼ 0.036 (c) For colony formation assay, DKK4 and vector-transfected stable cells were seeded in 6-well plates and colonies were visualized by crystal violet staining after 14 days. Fewer numbers of colonies were found in DKK4-expressing stable cells. *Po0.05. (d) Cell migration was assessed by wound-healing assay. DKK4 and vector-transfected stable cells were grown to confluency in cell monolayer after which a wound was made and cell migration was monitored at 0, 24 and 48 h. DKK4-expressing stable cells migrated slower towards the wound when compared with vector-transfected stable cells. Wound sizes were measured at each time point to calculate the percentage of wound closure, using the formula below, for vector and DKK4-transfected stable cells. *Po0.05, (T0hrÀT24 hr or 48 hr/T0hr) Â 100.

Oncogene Action of DKK4 on Wnt/b-catenin pathway S Fatima et al 4237

Oncogene Action of DKK4 on Wnt/b-catenin pathway S Fatima et al 4238

Figure 4 DKK4 reduces tumourigenicity, and b-catenin and cyclin D1 protein levels in vivo.(a) 97L_vec and 97L_DKK4 cells were s.c. injected into BALB/c nude mice, and tumours were monitored every week. Each point represents the mean tumour size of 8 mice±s.e.m. 97L_DKK4 cells formed smaller tumours when compared with control cells. *Po0.05, **Po0.005; ***Po0.0005. (b) Reduced b-catenin and cyclin D1 mRNA in 97L_DKK4 tumour xenografts compared with 97L_vec tumour xenografts as measured by qPCR. *Po0.05 (c) Immunohistochemical staining of DKK4, b-catenin and cyclin D1 in 97L_vec and 97L_DKK4 tumour xenograft tissues. Tumour xenografts from 97L_DKK4 cells had less expression of b-catenin and cyclin D1. Original magnification, Â 200.

Non-tumour Tumour

DKK4

-catenin

Figure 5 Downregulation of DKK4 is associated with b-catenin accumulation in HCC. Immunohistochemical staining of DKK4 and b-catenin in a representative HCC tissue and its corresponding adjacent non-tumour liver tissue is shown. Original magnification, Â 200.

system (Figure 6b). As GSK3b is crucial in b-catenin in total GSK3b levels were observed, 97L_DKK4 cells phosphorylation and its eventual degradation, total showed a reduction in phosphorylated GSK3b (phos- GSK3b levels were also examined. Although no changes pho-GSK3b, inactive form of GSK3b) compared with

Oncogene Action of DKK4 on Wnt/b-catenin pathway S Fatima et al 4239 an inhibitor of GSK3b activity. As shown in Figure 6c, incubation with lithium chloride blocked the ability of DKK4 to reduce b-catenin levels. Nevertheless, it is also likely that the effect of DKK4 on b-catenin phosphoryla- tion is mediated through GSK-3a as well as GSK-3b,as lithium chloride affects both forms equivalently and b-catenin is subjected to equal control by these two isoforms.

Discussion

DKK4 has been reported as a negative regulator of Wnt/ b-catenin signalling pathway (Krupnik et al.,1999)butits role in HCC is unknown. To our knowledge, the present study is the first to demonstrate reduced expression of DKK4 in HCC cell lines as well as clinical samples and its suppressive effect on tumourigenesis by influencing the Wnt/b-catenin pathway. We found DKK4 was not detectable in HCC cell lines (albeit to a low level in HepG2, a hepatic adenoma cell line) and that reduced expression was detected in 47% of HCC tissues compared with their corresponding non-tumour liver tissues. More- over, we propose DKK4 as a negative regulator of HCC cell growth as demonstrated by the functional studies. Ectopic expression of DKK4 in malignant HCC cells resulted in inhibition of cell growth, colony formation and cell migration in vitro as well as reduction of in vivo tumourigenicity in HCC xenograft mouse model. This is also the first study to demonstrate reduced expression of DKK4 relating to b-catenin accumulation in HCC tissues. Despite the potential tumour suppressive role shown in animal and in vitro studies, DKK4 expression in clinical samples was not significantly associated with any clinicopathological parameters studied. This is in agree- ment with a recent study on renal cell carcinoma, which did not find any clinical association with DKK4 expres- sion (Hirata et al., 2011). Conceivably, there may be other confounding factors (for example, oncogenic mutations of b-catenin) affecting the DKK4 tumour suppression functions in clinical tissues that deems necessary for further investigation. Epigenetic inactivation by DNA methylation is an important mechanism of gene silencing in human Figure 6 DKK4 affects Wnt/b-catenin pathway components and cancers. However, such changes have been reported degrades b-catenin by the proteasome. (a) Western blot analysis of for all DKK family members with the exception of proteins associated with the Wnt/b-catenin pathway in 97L_vec DKK4 (Sato et al., 2007; Hirata et al., 2009; Yang et al., and 97L_DKK4 cells. Expression of DKK4 in 97L stable cells 2010). Frequent downregulation of DKK1-DKK3 have reduced phospho-GSK3b and active b-catenin, whereas this induced the expression of phospho-b-catenin. (b) 97L_vec and been reported in gastrointestinal and colorectal cancers 97L_DKK4 cells were treated with 10 mM MG132 for 12 h, and b- because of promoter methylation, and Yang et al. (2010) catenin protein levels were detected by western blot. Inhibition of reported methylation in DKK2 and DKK3 in HCC the proteasome reduced the effect of DKK4 on b-catenin. Western (Aguilera et al., 2006; Sato et al., 2007). However, our blot bands were quantified by Image J Software, following result from HCC cell lines suggests that hypermethyla- normalization to b-actin. (c) 97L_vec and 97L_DKK4 cells were treated with 20 mM lithium chloride (LiCl), for 12 h and b-catenin, tion does not seem to be responsible for the down- phospho-GSK3b and total GSK3b protein levels were analysed by regulation of DKK4 in HCC. Similarly, Baehs et al. western blot. The effect of DKK4 on b-catenin was masked by (2009) did not report any CpG islands in DKK4 treatment with LiCl. promoter in colorectal cancer. This suggests that reasons other than methylation could be responsible for the 97L_vec cells. To examine whether reduction of b-catenin reduced expression of DKK4. One possible reason for by DKK4 requires the activity of GSK3b, 97L_vec cells the reduced DKK4 expression could be because of allelic and 97L_DKK4 cells were treated with lithium chloride, imbalance. DKK4 is located on chromosome 8p11.2–p11.1,

Oncogene Action of DKK4 on Wnt/b-catenin pathway S Fatima et al 4240 and this chromosome experiences frequent loss of hetero- As a negative regulator and downstream target of Wnt/ zygosity in many cancers including HCC (Pineau et al., b-catenin signalling, DKK4 like DKK1 provides a 1999). Chromosome 8p is also home to several candidate negative feedback loop that is important for regulating tumour suppressor genes including deleted in liver cancer 1 Wnt/b-catenin signalling in normal tissues (Krupnik (Yuan et al., 1998), fasciculation and elongation protein et al., 1999; Pendas-Franco et al., 2008). Thus, loss of zeta-1 (Ishii et al., 1999) and lipid transfer proteins (Liao DKK4 expression in HCC may provide cancer cells with et al., 2000). The commonly deleted regions in HCC have a growth advantage. However, overexpression of DKK4 been identified at 8p21.3–p22 and 8p23, but the recent in colon cancer and renal cell carcinoma suggests that identification of 8p11.2 as a region harbouring metastatic DKK4 may have tissue-specific functions or may serve tumour suppressor genes (Zhang et al., 2003) makes it a roles other than regulating the Wnt/b-catenin pathway. likely candidate region to study DKK4 allelic imbalance. Cytoplasmic and/or nuclear accumulation of b-catenin Thus, further investigation is warranted to understand the is a hallmark of aberrant Wnt/b-catenin signalling, and mechanisms that lead to DKK4 inactivation. this is the first study to suggest that inactivation of DKK4 Our results show that DKK4 reduces only cytoplas- may influence the Wnt/b-catenin pathway in HCC. HCC mic b-catenin levels and does not affect nuclear b- clinical samples with b-catenin accumulation showed catenin levels in vitro. This suggests that DKK4 may not reduced DKK4 expression. Reduced expression of b- be sufficient to suppress the Wnt/b-catenin pathway on catenin and cyclin D1 were also found in tumour its own and may require additional suppressive compo- xenografts with high DKK4 expression. Interestingly, nents to attenuate Wnt/b-catenin signalling. Krieghoff Matsui et al. (2009) recently associated upregulation of et al. (2006) reported that b-catenin localization in the DKK4 to b-catenin accumulation in colorectal cancer. cytoplasm and its subsequent translocation to the Our results also show that HCC stable cell line over- nucleus could be governed by competition between expressing DKK4 has unaltered levels of total GSK3b but cytoplasmic and nuclear retention factors (for example, shows reduced levels of phospho-GSK3b (inactive BCL9, APC, Axin and TCF4) for b-catenin. DKK4 also GSK3b). GSK3b is a protein kinase associated with the contains TCF-binding sites (Baehs et al., 2009) and destruction complex for b-catenin degradation. This several studies have demonstrated other antagonists of suggests that active GSK3b is important for b-catenin the Wnt/b-catenin pathway to compete with the TCF phosphorylation and DKK4-mediated b-catenin reduc- family in the nucleus to attenuate Wnt/b-catenin tion. This is also supported by the luciferase assay, which signalling (Takemaru et al., 2003; Ito et al., 2008; Wu shows that DKK4 reduces TCF-mediated luciferase et al., 2010). Several secretory proteins including DKK3 activity of wild-type b-catenin but has no effect on b- have also been reported in the nucleus (Lee et al., 2009a; catenin containing mutation at Ser45. Mutation at Ser45 Wiesman et al., 2010; Zhang et al., 2010). Most recently, renders GSK3b unable to phosphorylate b-catenin at Liao et al. (2011) demonstrated positive regulation of Ser33/Ser37/Thr41, thus leading to b-catenin stabilization DKK4 by thyroid hormone receptor in promoting (Sakanaka, 2002). Additionally, treatment of HCC cells tumour suppression in HCC, further suggesting DKK4 overexpressing DKK4 with MG132, a proteasome to be influenced by other signalling cascades. inhibitor, masked the effect of DKK4 on b-catenin. This Several studies have shown varied expression and is a novel mechanism identified in HCC. Further studies function of DKK4 in different cancer types. Baehs et al. are warranted to understand the molecular mechanism (2009) reported downregulation of DKK4 in 12 of 21 leading to DKK4-mediated b-catenin reduction. Yama- colorectal cancers and overexpression of DKK4 inhib- moto et al. (2008) recently demonstrated a clathrin- ited colorectal cell growth. However, this is contrary to dependent internalization of LRP6 for DKK1 inhibition the findings by Pendas-Franco et al. (2008) and Matsui of Wnt/b-catenin signalling. However, contrary to results et al. (2009) who reported overexpression of DKK4 in of Yamamoto et al. (2008), Blitzer et al. (2006) used colon cancer. Although the reason for this discrepancy is mouse fibroblast cells to show that clathrin-dependent unclear, the expression of DKK4 may be reflective of internalization of Wnt is required to propagate Wnt/b- expression of other genes in colon cancer. Pendas- catenin signalling and disturbing the clathrin-mediated Franco et al. (2008) associated overexpression of DKK4 endocytosisblocksWnt/b-catenin activity. These recent in all cases of colon cancers studied to reduced findings suggest that mechanisms of DKK antagonistic expression of vitamin D receptor that resulted in activity may vary in different cell types. angiogenesis and promoted cell invasiveness. Matsui Among the several antagonists of the Wnt/b-catenin et al. (2009) correlated high DKK4 expression to pathway that have been identified, including Cerberus, fibroblast growth factor-20. Hirata et al. (2011) also DKKs, secreted frizzled-related protein (SFRP), SOST, observed overexpression of DKK4 in 19 out of 30 renal Wnt inhibitory factor-1 (WIF-1) and Wise, only sFRP, cancer tissues, and reported greater invasiveness and WIF-1 and DKK1 have been reported to attenuate migration in renal carcinoma-derived Caki cells Wnt/b-catenin signalling in HCC (Shih et al., 2007; overexpressing DKK4. Furthermore, although ectopic Hu et al., 2009; Yu et al., 2009). Additionally, we and overexpression of DKK4 reduced b-catenin protein others have shown a correlation between an increase in levels in Caki cells, it activated the JNK pathway and DKK1 serum and transcript level in HCC with b-catenin promoted tumour growth in vivo. This suggests that accumulation resulting in poor prognosis (Yu et al., 2009; DKK4 activates the non-canonical JNK pathway while Tung et al., 2011). Being an antagonist of Wnt/b-catenin inhibiting the Wnt/b-catenin pathway in renal cancer. signalling and also a downstream target of b-catenin, the

Oncogene Action of DKK4 on Wnt/b-catenin pathway S Fatima et al 4241 overexpression of DKK1 has been proposed to be a result Table 2. Tissue samples were snap-frozen in liquid nitrogen of dysregulated Wnt/b-catenin signalling in HCC or due immediately after resection and stored in À80 1C until used, as to an abrogated negative feedback mechanism (Yu et al., described previously (Yi et al., 2008). 2009). Few studies have associated frequent methylation of DKK2 and DKK3 in HCC to poor prognosis (Ding Real-time qPCR et al., 2009; Yang et al., 2010) implying their roles as Total RNA was isolated using TRIzol reagent (Invitrogen) prognostic markers. However, contrary to DKK1 and and the residual genomic DNA was removed by treating the DKK4, which affect b-catenin accumulation, DKK2 and RNA samples with DNase I (Applied Biosystems, Foster City, DKK3 have not been studied in the context of b-catenin CA, USA). RNA pellets were then dissolved in diethylpyr- upregulation in HCC and furthermore, DKK3 neither ocarbonate-treated water and stored at À80 1C before use. affects TCF-mediated luciferase activity nor does ectopic Reverse transcription and qPCR were performed as described previously (Lee et al., 2008). Total RNA (1 mg) was reverse overexpression of DKK3 in immortalized normal hepa- transcribed into cDNA using TaqMan Reverse Transcription tocytes, (MIHA cell line), affect b-catenin protein levels Reagents (Applied Biosystems). qPCR was performed using in vitro (Fatima S et al, unpublished data). Thus, despite SYBR Green PCR Master Mix (Invitrogen) in a 7900HT of belonging to the same family, different DKKs may Sequence Detection Systems (Applied Biosystems) using serve different functions in HCC. Other antagonists of primers for DKK4 (forward: 50-CCA GAA AGT TCT GCC Wnt, including WIF-1 (Ding et al., 2009) and SFRPs TCC AG-30 and reverse: 50-CTT CTG CAT GTG TGC CAT (Takagi et al., 2008) are epigenetically silenced in HCC, CT-30). All reactions were performed in duplicates under the and their association to b-catenin upregulation in HCC following cycling conditions: 50 1C for 2 min, 95 1C for 10 min, has not been reported. Although different Wnt antago- 40 cycles of 95 1C for 15 s and 60 1C for 1 min. The relative nists have different expression and clinical association in mRNA expression was normalized using glyceraldehyde 3-phosphate dehydrogenase as an internal control. 2–(delta) HCC, the combined expression of some of these (delta)Ct method was used to calculate the fold change in gene antagonists may give a better understanding of their expression between HCC tissues and their corresponding non- prognostic significance in HCC. Recently, Urakami et al. tumour liver tissues, as described previously (Yi et al., 2008). (2006) reported that combined methylation analysis of HCC tissue with DKK4 expression less than twofold of the several Wnt antagonists (DKK3, WIF-1 and sFRPs) can non-tumour liver tissue was defined as negative for DKK4 serve as tumour marker for diagnosis and prognosis for expression (DKK4À), and those with more than twofold renal cell carcinoma. Further work is needed to delineate expression of the non-tumour liver tissue was defined as the mechanisms for the differing expression of DKKs in positive for DKK4 expression (DKK4 þ ). HCC and also to assess their tumour suppressive (or oncogenic) roles in hepatocarcinogenesis. 5-azacytidine treatment Altogether, we have shown reduced expression of HCC cells were seeded in a 6-well plate 24 h before treatment. DKK4 in HCC and its tumour suppressive properties Cells were then treated with 0 mM,5mM and 10 mM 5-azacytidine both in vitro and in vivo, suggesting its role as an (Sigma-Aldrich, St Louis, MO, USA) for 48 h. Drugs and important regulator of HCC. Although several studies culture medium were refreshed daily. have highlighted the tumourigenic and prognostic significance of aberrant Wnt/b-catenin signalling in Dual luciferase assay HCC, the reasons for its aberrant activation require Dual luciferase assay was performed as previously described (Liu further mechanistic investigations. DKK4 may serve as et al., 2009). Briefly, cells were co-transfected, using Lipofecta- a promising target for cancer treatment, though detailed mine 2000 (Invitrogen), with TCF Reporter Plasmid (TOPflash) study is necessary to delineate how this antagonist exerts (Millipore, Billercia, MA, USA) or those with mutant TCF- binding sites (FOPflash) (Millipore) along with the transfection its antitumour effects. control Renilla Luciferase Reporter Vector (pRL-TK) (Promega, Madison, WI, USA). In some experiments, cells were co- Materials and methods transfected with DKK4 and wild-type b-catenin (Morin et al., 1997) (Addgene plasmid 16518) or mutant b-catenin (Morin et al., 1997) (with a 3 bp deletion at amino-acid Ser45) (Addgene Human liver cell lines and clinical tissues plasmid 16520) (Addgene, Cambridge, MA, USA). At 48 h after Immortalized human hepatocytes (MIHA) and eight HCC cell transfection, cells were collected and suspended in Reporter lines, HepG2, Hep3B, HuH-7, PLC, H2P, H2M, 97L and Lysis Buffer (Promega) and luciferase activities were measured MHCC97H (97H), were cultured in Dulbecco’s modified Eagle using reagents in Dual-Luciferase Reporter Assay System medium (Invitrogen, Carlsbad, CA, USA) supplemented with (Promega) and a luminometer. 10% foetal bovine serum (Invitrogen) and 1% penicillin and streptomycin (Invitrogen), as described previously (Liu et al., 2009; Lee et al., 2010). Cultures were maintained in a Preparation of cytoplasmic and nuclear extracts humidified incubator at 37 1C with 5% carbon dioxide. Cytoplasmic and nuclear extracts were prepared as described A total of 81 pairs of tumour and adjacent non-tumour liver previously (Dignam et al., 1983). Briefly, confluent cells from tissues were obtained from patients with primary HCC 60 mm dishes were washed twice with cold phosphate-buffered resected at Queen Mary Hospital, Hong Kong between saline and resuspended in buffer A (10 mM HEPES (4-(2- 1993–2005. Consent regarding the use of clinical specimens hydroxyethyl)-1-piperazineethanesulfonic acid), pH 7.9, 1.5 mM for this study was obtained from Institutional Review Board of MgCl2,10mM KCl, 0.5 mM dithiothreitol and 0.5 mM phenyl- the University of Hong Kong/Hospital Authority Hong Kong methanesulfonylfluoride) with 0.3% Nonidet P-40, incubated West Cluster. Demographic data and clinicopathological on ice for 15 min and centrifuged at 2000 g for 10 min at 4 1C. features of the patients are summarized in Supplementary The supernatant was collected, which represented the cyto-

Oncogene Action of DKK4 on Wnt/b-catenin pathway S Fatima et al 4242 plasmic extract. The nuclear pellet was resuspended in buffer C either with 20 mM lithium chloride (USB, Cleveland, OH, (20 mM HEPES, pH 7.9, 25% (v/v) glycerol, 420 mM NaCl, USA) or 10 mM MG132 (Sigma-Aldrich) for 12 h. Protein 1.5 mM MgCl2, 0.2 mM EDTA, 0.5 mM dithiothreitol and lysates were collected for western blot as described above. 0.5 mM phenylmethanesulfonylfluoride), incubated on ice for 30 min with occasional mixing and centrifuged at 20 000 g for HCC xenograft mouse model 15 min. The supernatant represented the nuclear extract. The Male athymic BALB/c nude mice were bred in the Laboratory cytoplasmic and nuclear extracts were stored at À80 1C until Animal Unit, the University of Hong Kong. Institutional used for subsequent western blot analysis. guidelines were followed while handling the animals. The animals were placed in a pathogen-free environment and Western blot allowed to acclimate for 1 week before being used in the study. Cell lysates were extracted using radioimmunoprecipitation A total of 1 Â 106 97L_vec cells or 97L_DKK4 cells suspended assay buffer (150 mM sodium chloride, 1% Triton X-100, 0.5% in phosphate-buffered saline (200 ml per mouse) were injected 3 sodium deoxycholate, 0.1% SDS, 50 mM Tris, pH 8.0) and s.c. into 4-week-old nude mice. Tumour volume in mm was protein levels were detected following the procedure as measured weekly with digital calliper using the formula: described previously (Lu et al., 2011). b-Actin was used as a volume ¼ (width)2 Â length/2. Animals were killed 8 weeks loading control for each sample. The antibodies used for after s.c. injection. The protocol was approved by the western blot, along with their dilutions, included b-actin Committee on the Use of Live Animals in Teaching and (1:5000, Sigma-Aldrich), b-catenin (1:1000, Cell Signaling Research, the University of Hong Kong. Technology, Danvers, MA, USA), active b-catenin (1:1000, Millipore), phosphorylated b-catenin (Thr41/Ser45) (1:1000, Immunohistochemistry Cell Signaling Technology), total GSK3b (1:1000, Cell Immunohistochemistry was performed as previously described Signaling Technology), phosphorylated GSK3b (Ser9) with modifications (Lee et al., 2006, 2009b). Tissue sections were (1:1000, Cell Signaling Technology), cyclin D1 (1:1000, deparaffinized by submerging the slides in xylene and rehydrated Invitrogen) and DKK4 (1:250, Abgent, San Diego, CA, USA). in decreasing concentrations of ethanol (100% followed by 90 and 75%, each twice). Antigen retrieval was achieved by boiling the tissue sections in EDTA buffer (1 mM EDTA, 0.05% Tween Establishment of cell clones overexpressing DKK4 20, pH 8.0) in a microwave for 10 min before incubation with b- DKK4 expression construct used here was described in catenin (1:100, BD, Franklin Lakes, NJ, USA), DKK4 (1:200, previous study (Baehs et al., 2009). PLC and 97L cell lines Abgent) or cyclin D1 (1:100, Cell Signaling Technology) were transfected with: (a) pcDNA3.1(À) vector only (PLC_vec antibodies overnight. Sections were then incubated for 1 h with and 97L_vec) and (b) DKK4 (PLC_DKK4 and 97L_DKK4) horseradish peroxidase-conjugated secondary antibodies (Invi- using Lipofectamine 2000 (Invitrogen). Transfected cells were trogen), and diaminobenzidine (Dako, Glostrup, Denmark) was grown in Dulbecco’s modified Eagle medium with 10% foetal used as the chromogen. The tissue sections were scored for bovine serum with addition of 500 mg/ml G418 antibiotic intensity using the criteria: 0 (no signal), 1 (weak; 10–20% (Invitrogen) for 3 weeks. For transient transfection experi- positive), 2 (moderate; 20–50% positive) and 3 (strong; 450% ment, the methodology of transfection was the same, except positive). that no drug selection using G418 was performed. Over- expression of DKK4 was confirmed using western blot. DKK4 antibody specificity For antibody specificity by western blot and immunohistochem- MTT cell proliferation assay ical analysis, DKK4-blocking peptide (Abgent Inc.; catalogue # PLC and 97L cells stably transfected with DKK4 or vector BP1524a) was used at a 10-fold molar excess to DKK4 antibody were seeded in 96-well plates at an initial cell density of 2000 (Abgent), as per the manufacturer’s instructions. Western blot cells per well in triplicates. Cell viability was assessed by the (Lu et al., 2011) and immunohistochemical analysis (Lee et al., reduction of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazo- 2006, 2009b) were performed as described previously. lium bromide (MTT) (Sigma-Aldrich) as previously described (Liu et al., 2009). Briefly, 10 ml MTT (5 mg/ml) was added to Statistical analysis each well and plates were incubated at 37 1C for 3 h. After The data were analysed using Prism for Windows (version 5.01) incubation, dimethyl sulfoxide was added and the colour (GraphPad Software, La Jolla, CA, USA) and SPSS for formation was quantified by means of an enzyme-linked Windows (version 16.0) (IBM, New York, NY, USA). Student’s immunosorbent assay plate reader at 570 nm wavelength. t-tests were used to study significant differences between different groups for in vitro and in vivo functional experiments. For clinicopathological analysis, Pearson’s w2 was used to Wound-healing assay calculate the P-value (P). Po 0.05 was considered statistically PLC and 97L cells stably transfected with DKK4 or vector significant. were grown to 90% confluence in 6-well plates following the procedure as previously described (Liu et al., 2010). Conflict of interest Colony-formation assay PLC and 97 L cells stably transfected with DKK4 and vector The authors declare no conflict of interest. were seeded in 6-well plates at an initial cell density of 200 cells per well. Colonies were allowed to form for 2 weeks and stained with 0.05% crystal violet after methanol fixation. Acknowledgements

Lithium chloride and MG132 treatment We thank Dr Bhavin M Thakkar of the National University of In all, 97L cells stably transfected with DKK4 or vector were Singapore, Singapore for his assistance in immunohistochemical grown to 90% confluence in 6-well plates and were treated studies.

Oncogene Action of DKK4 on Wnt/b-catenin pathway S Fatima et al 4243 References

Aguilera O, Fraga MF, Ballestar E, Paz MF, Herranz M, Espada J Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. (2011). et al. (2006). Epigenetic inactivation of the Wnt antagonist Global cancer statistics. CA Cancer J Clin 61: 69–90. DICKKOPF-1 (DKK-1) gene in human colorectal cancer. Kemp W, Pianko S, Nguyen S, Bailey MJ, Roberts SK. (2005). Oncogene 25: 4116–4121. Survival in hepatocellular carcinoma: impact of screening and Apte U, Thompson MD, Cui S, Liu B, Cieply B, Monga SP. (2008). etiology of liver disease. J Gastroenterol Hepatol 20: 873–881. Wnt/beta-catenin signaling mediates oval cell response in rodents. Krieghoff E, Behrens J, Mayr B. (2006). Nucleo-cytoplasmic Hepatology 47: 288–295. distribution of beta-catenin is regulated by retention. J Cell Sci Baehs S, Herbst A, Thieme SE, Perschl C, Behrens A, Scheel S et al. 119: 1453–1463. (2009). Dickkopf-4 is frequently down-regulated and inhibits Krupnik VE, Sharp JD, Jiang C, Robison K, Chickering TW, growth of colorectal cancer cells. Cancer Lett 276: 152–159. Amaravadi L et al. (1999). Functional and structural diversity of Blitzer JT, Nusse R. (2006). A critical role for endocytosis in Wnt the human Dickkopf gene family. Gene 238: 301–313. signaling. BMC Cell Biol 7: 28. Lee EJ, Jo M, Rho SB, Park K, Yoo YN, Park J et al. (2009a). Dkk3, Chamorro MN, Schwartz DR, Vonica A, Brivanlou AH, Cho KR, downregulated in cervical cancer, functions as a negative regulator Varmus HE. (2005). FGF-20 and DKK1 are transcriptional targets of beta-catenin. Int J Cancer 124: 287–297. of beta-catenin and FGF-20 is implicated in cancer and develop- Lee NP, Chen L, Lin MC, Tsang FH, Yeung C, Poon RT et al. ment. EMBO J 24: 73–84. (2009b). Proteomic expression signature distinguishes cancerous and Chen TC, Hsieh LL, Ng KF, Jeng LB, Chen MF. (1998). Absence of nonmalignant tissues in hepatocellular carcinoma. J Proteome Res APC gene mutation in the mutation cluster region in hepatocellular 8: 1293–1303. carcinoma. Cancer Lett 134: 23–28. Lee NP, Leung KW, Cheung N, Lam BY, Xu MZ, Sham PC et al. de La Coste A, Romagnolo B, Billuart P, Renard CA, Buendia MA, (2008). Comparative proteomic analysis of mouse livers from Soubrane O et al. (1998). Somatic mutations of the beta-catenin embryo to adult reveals an association with progression of gene are frequent in mouse and human hepatocellular carcinomas. hepatocellular carcinoma. Proteomics 8: 2136–2149. Proc Natl Acad Sci U S A 95: 8847–8851. Lee NP, Tong MK, Leung PP, Chan VW, Leung S, Tam PC et al. Dignam JD, Lebovitz RM, Roeder RG. (1983). Accurate transcription (2006). Kidney claudin-19: localization in distal tubules and initiation by RNA polymerase II in a soluble extract from isolated collecting ducts and dysregulation in polycystic renal disease. FEBS mammalian nuclei. Nucleic Acids Res 11: 1475–1489. Lett 580: 923–931. Ding Z, Qian YB, Zhu LX, Xiong QR. (2009). Promoter methylation Lee NP, Tsang FH, Shek FH, Mao M, Dai H, Zhang C et al. (2010). and mRNA expression of DKK-3 and WIF-1 in hepatocellular Prognostic significance and therapeutic potential of eukaryotic carcinoma. World J Gastroenterol 15: 2595–2601. translation initiation factor 5A (eIF5A) in hepatocellular carcino- Fatima S, Lee NP, Luk JM. (2011). Dickkopfs and Wnt/beta-catenin ma. Int J Cancer 127: 968–976. signalling in liver cancer. World J Clin Oncol 2: 311–325. Liao C, Zhao M, Song H, Uchida K, Yokoyama KK, Li TP. (2000). Gonzalez FJ. (2006). Role of beta-catenin in the adult liver. Identification of the gene for a novel liver-related putative tumor Hepatology 43: 650–653. suppressor at a high-frequency loss of heterozygosity region of Gonzalez-Sancho JM, Aguilera O, Garcia JM, Pendas-Franco N, Pena chromosome 8p23 in human hepatocellular carcinoma. Hepatology C, Cal S et al. (2005). The Wnt antagonist DICKKOPF-1 gene is a 32: 721–727. downstream target of beta-catenin/TCF and is downregulated in Liao CH, Yeh CT, Huang YH, Wu SM, Chi HC, Tsai MM et al. human colon cancer. Oncogene 24: 1098–1103. (2011). Dickkopf 4 positively regulated by the thyroid hormone Hao K, Luk JM, Lee NP, Mao M, Zhang C, Ferguson MD et al. receptor suppresses cell invasion in human hepatoma cells. Hepatology. (2009). Predicting prognosis in hepatocellular carcinoma after Liu AM, Poon RT, Luk JM. (2010). MicroRNA-375 targets Hippo- curative surgery with common clinicopathologic parameters. BMC signaling effector YAP in liver cancer and inhibits tumor properties. Cancer 9: 389. Biochem Biophys Res Commun 394: 623–627. Hirata H, Hinoda Y, Majid S, Chen Y, Zaman MS, Ueno K et al. Liu LX, Lee NP, Chan VW, Xue W, Zender L, Zhang C et al. (2009). (2011). DICKKOPF-4 activates the noncanonical c-Jun-NH2 Targeting cadherin-17 inactivates Wnt signaling and inhibits tumor kinase signaling pathway while inhibiting the Wnt-canonical growth in liver carcinoma. Hepatology 50: 1453–1463. pathway in human renal cell carcinoma. Cancer 117: 1649–1660. Logan CY, Nusse R. (2004). The Wnt signaling pathway in Hirata H, Hinoda Y, Nakajima K, Kawamoto K, Kikuno N, development and disease. Annu Rev Cell Dev Biol 20: 781–810. Kawakami K et al. (2009). Wnt antagonist gene DKK2 is Lu WJ, Lee NP, Kaul SC, Lan F, Poon RT, Wadhwa R et al. epigenetically silenced and inhibits renal cancer progression through (2011). Induction of mutant p53-dependent apoptosis in human apoptotic and cell cycle pathways. Clin Cancer Res 15: 5678–5687. hepatocellular carcinoma by targeting stress protein mortalin. Int J Hu J, Dong A, Fernandez-Ruiz V, Shan J, Kawa M, Martinez-Anso E Cancer 129: 1806–1814. et al. (2009). Blockade of Wnt signaling inhibits angiogenesis MacDonald BT, Tamai K, He X. (2009). Wnt/beta-catenin and tumor growth in hepatocellular carcinoma. Cancer Res 69: signaling: components, mechanisms, and diseases. Dev Cell 17: 6951–6959. 9–26. Huang H, Fujii H, Sankila A, Mahler-Araujo BM, Matsuda M, Maehata T, Taniguchi H, Yamamoto H, Nosho K, Adachi Y, Cathomas G et al. (1999). Beta-catenin mutations are frequent in Miyamoto N et al. (2008). Transcriptional silencing of Dickkopf human hepatocellular carcinomas associated with hepatitis C virus gene family by CpG island hypermethylation in human infection. Am J Pathol 155: 1795–1801. gastrointestinal cancer. World J Gastroenterol 14: 2702–2714. Inagawa S, Itabashi M, Adachi S, Kawamoto T, Hori M, Shimazaki J Mao B, Niehrs C. (2003). Kremen2 modulates Dickkopf2 activity et al. (2002). Expression and prognostic roles of beta-catenin in during Wnt/LRP6 signaling. Gene 302: 179–183. hepatocellular carcinoma: correlation with tumor progression and Mao B, Wu W, Davidson G, Marhold J, Li M, Mechler BM et al. postoperative survival. Clin Cancer Res 8: 450–456. (2002). Kremen proteins are Dickkopf receptors that regulate Wnt/ Ishii H, Baffa R, Numata SI, Murakumo Y, Rattan S, Inoue H et al. beta-catenin signalling. Nature 417: 664–667. (1999). The FEZ1 gene at chromosome 8p22 encodes a leucine- Mao BY, Wu W, Li Y, Hoppe D, Stannek P, Glinka A et al. (2001). zipper protein, and its expression is altered in multiple human LDL-receptor-related protein 6 is a receptor for Dickkopf proteins. tumors. Proc Natl Acad Sci U S A 96: 3928–3933. Nature 411: 321–325. Ito K, Lim AC, Salto-Tellez M, Motoda L, Osato M, Chuang LS et al. Matsui A, Yamaguchi T, Maekawa S, Miyazaki C, Takano S, Uetake (2008). RUNX3 attenuates beta-catenin/T cell factors in intestinal T et al. (2009). DICKKOPF-4 and -2 genes are upregulated in tumorigenesis. Cancer Cell 14: 226–237. human colorectal cancer. Cancer Sci 100: 1923–1930.

Oncogene Action of DKK4 on Wnt/b-catenin pathway S Fatima et al 4244 Morin PJ, Sparks AB, Korinek V, Barker N, Clevers H, Vogelstein B and tissue Dickkopf-1 levels in human hepatocellular carcinoma. et al. (1997). Activation of beta-catenin-Tcf signaling in Liver Int 31: 1494–1504. colon cancer by mutations in beta-catenin or APC. Science 275: Urakami S, Shiina H, Enokida H, Hirata H, Kawamoto K, Kawakami 1787–1790. T et al. (2006). Wnt antagonist family genes as biomarkers for Nakamura T, Matsumoto K. (2008). The functions and possible diagnosis, staging, and prognosis of renal cell carcinoma using significance of Kremen as the gatekeeper of Wnt signalling in tumor and serum DNA. Clin Cancer Res 12: 6989–6997. development and pathology. J Cell Mol Med 12: 391–408. Wiesman KC, Wei L, Baughman C, Russo J, Gray MR, Castellot JJ. Nejak-Bowen KN, Monga SP. (2011). Beta-catenin signaling, liver (2010). CCN5, a secreted protein, localizes to the nucleus. J Cell regeneration and hepatocellular cancer: sorting the good from the Commun Signal 4: 91–98. bad. Semin Cancer Biol 21: 44–58. Wong CM, Fan ST, Ng IO. (2001). beta-Catenin mutation and Niehrs C. (2006). Function and biological roles of the Dickkopf family overexpression in hepatocellular carcinoma: clinicopathologic and of Wnt modulators. Oncogene 25: 7469–7481. prognostic significance. Cancer 92: 136–145. Niida A, Hiroko T, Kasai M, Furukawa Y, Nakamura Y, Suzuki Y Wu Y, Zhang Y, Zhang H, Yang X, Wang Y, Ren F et al. et al. (2004). DKK1, a negative regulator of Wnt signaling, is a (2010). p15RS attenuates Wnt/{beta}-catenin signaling by target of the beta-catenin/TCF pathway. Oncogene 23: 8520–8526. disrupting {beta}-catenin.TCF4 Interaction. J Biol Chem 285: Orford K, Crockett C, Jensen JP, Weissman AM, Byers SW. (1997). 34621–34631. Serine phosphorylation-regulated ubiquitination and degradation of Yang B, Du Z, Gao YT, Lou C, Zhang SG, Bai T et al. (2010). beta-catenin. J Biol Chem 272: 24735–24738. Methylation of Dickkopf-3 as a prognostic factor in cirrhosis- Pendas-Franco N, Garcia JM, Pena C, Valle N, Palmer HG, related hepatocellular carcinoma. World J Gastroenterol 16: 755–763. Heinaniemi M et al. (2008). DICKKOPF-4 is induced by TCF/ Yamamoto H, Sakane H, Michiue T, Kikuchi A. (2008). Wnt3a and beta-catenin and upregulated in human colon cancer, promotes Dkk1 regulate distinct internalization pathways of LRP6 to tune the tumour cell invasion and angiogenesis and is repressed by activation of beta-catenin signaling. Dev Cell 15: 37–48. 1alpha,25-dihydroxyvitamin D3. Oncogene 27: 4467–4477. Yi X, Luk JM, Lee NP, Peng J, Leng X, Guan XY et al. (2008). et al Pineau P, Nagai H, Prigent S, Wei Y, Gyapay G, Weissenbach J . Association of mortalin (HSPA9) with liver cancer metastasis (1999). Identification of three distinct regions of allelic deletions on and prediction for early tumor recurrence. Mol Cell Proteomics 7: the short arm of chromosome 8 in hepatocellular carcinoma. 315–325. Oncogene 18: 3127–3134. Yu B, Yang X, Xu Y, Yao G, Shu H, Lin B et al. (2009). Elevated Sakanaka C. (2002). Phosphorylation and regulation of beta-catenin expression of DKK1 is associated with cytoplasmic/nuclear by casein kinase I epsilon. J Biochem 132: 697–703. beta-catenin accumulation and poor prognosis in hepatocellular Sato H, Suzuki H, Toyota M, Nojima M, Maruyama R, Sasaki S et al. carcinomas. J Hepatol 50: 948–957. (2007). Frequent epigenetic inactivation of DICKKOPF Yuan BZ, Miller MJ, Keck CL, Zimonjic DB, Thorgeirsson SS, family genes in human gastrointestinal tumors. Carcinogenesis 28: 2459–2466. Popescu NC. (1998). Cloning, characterization, and chromosomal Semenov MV, Tamai K, Brott BK, Kuhl M, Sokol S, He X. (2001). localization of a gene frequently deleted in human liver cancer Head inducer Dickkopf-1 is a ligand for Wnt coreceptor LRP6. (DLC-1) homologous to rat RhoGAP. Cancer Res 58: 2196–2199. Curr Biol 11: 951–961. Yue W, Sun Q, Dacic S, Landreneau RJ, Siegfried JM, Yu J et al. Shih YL, Hsieh CB, Lai HC, Yan MD, Hsieh TY, Chao YC et al. (2008). Downregulation of Dkk3 activates beta-catenin/TCF-4 (2007). SFRP1 suppressed hepatoma cells growth through Wnt signaling in lung cancer. Carcinogenesis 29: 84–92. canonical signaling pathway. Int J Cancer 121: 1028–1035. Zeng X, Tamai K, Doble B, Li S, Huang H, Habas R et al. (2005). A Takagi H, Sasaki S, Suzuki H, Toyota M, Maruyama R, Nojima M dual-kinase mechanism for Wnt co-receptor phosphorylation and et al. (2008). Frequent epigenetic inactivation of SFRP genes in activation. Nature 438: 873–877. hepatocellular carcinoma. J Gastroenterol 43: 378–389. Zhang K, Watanabe M, Kashiwakura Y, Li SA, Edamura K, Huang P Takemaru K, Yamaguchi S, Lee YS, Zhang Y, Carthew RW, Moon et al. (2010). Expression pattern of REIC/Dkk-3 in various cell RT. (2003). Chibby, a nuclear beta-catenin-associated antagonist of types and the implications of the soluble form in prostatic acinar the Wnt/Wingless pathway. Nature 422: 905–909. development. Int J Oncol 37: 1495–1501. Taniguchi K, Roberts LR, Aderca IN, Dong X, Qian C, Murphy LM Zhang LH, Qin LX, Ma ZC, Ye SL, Liu YK, Ye QH et al. et al. (2002). Mutational spectrum of beta-catenin, AXIN1, and (2003). Allelic imbalance regions on chromosomes 8p, 17p and 19p AXIN2 in hepatocellular carcinomas and hepatoblastomas. Onco- related to metastasis of hepatocellular carcinoma: comparison gene 21: 4863–4871. between matched primary and metastatic lesions in 22 patients by Tung EK, Mak CK, Fatima S, Lo RC, Zhao H, Zhang C et al. genome-wide microsatellite analysis. J Cancer Res Clin Oncol 129: (2011). Clinicopathological and prognostic significance of serum 279–286.

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

Oncogene