Transcription factor AP-4 promotes tumorigenic capability and activates the Wnt/β-catenin pathway in hepatocellular carcinoma.

It has been reported that the transcription factor activating enhancer-binding protein 4 (TFAP4) is upregulated and associated with an aggressive phenotype in several cancers. However, the precise mechanisms underlying the oncogenic role of TFAP4 remain largely unknown.
Methods: TFAP4 expression levels in hepatocellular carcinoma (HCC) cells and tissues were detected by quantitative real-time PCR (qPCR) and immunohistochemistry (IHC). In vitro and in vivo assays were performed to investigate the oncogenic function of TFAP4 in the tumor-initiating cell (TIC)-like phenotype and the tumorigenic capability of HCC cells. Luciferase reporter and chromatin immunoprecipitation (ChIP)-qPCR assays were performed to determine the underlying mechanism of TFAP4-mediated HCC aggressiveness.
Results: TFAP4 was markedly upregulated in human HCC, and was associated with significantly poorer overall and relapse-free survival in patients with HCC. Furthermore, we found that overexpression of TFAP4 significantly enhanced, whereas silencing TFAP4 inhibited, the tumor sphere formation ability and proportion of side-population cells in HCC cells in vitro, and ectopic TFAP4 enhanced the tumorigenicity of HCC cells in vivo. Mechanistically, we demonstrated that TFAP4 played an important role in activating Wnt/β-catenin signaling by directly binding to the promoters of DVL1 (dishevelled segment polarity protein 1) and LEF1 (lymphoid enhancer binding factor 1). Conclusions: Our results provide new insight into the mechanisms underlying hyperactivation of the Wnt/β-catenin pathway in HCC, as well the oncogenic ability of TFAP4 to enhance the tumor-forming ability of HCC cells.

Introduction

Hepatocellular carcinoma (HCC) is the third most common cause of cancer-related deaths; the 5-year survival rate for patients who do not receive treatment is less than 5% 1. In China, patients with HCC account for nearly 50% of all new cases worldwide 2. Unfortunately, the post-resection recurrence rate at 2 years is 50% 3 and the 5-year recurrence rate for patients with intermediate or advanced HCC is as high as 74% 4. Recurrence is mainly attributed to the high tumorigenic potential and drug resistance of HCC cells 5. Interestingly, these features can be detected in a small subpopulation of cells in HCC tissues, termed tumor-initiating cells (TICs) 6-8. TICs have high self-renewal capability and are believed to give rise to tumor heterogeneity 9, 10. An increasing number of studies support the notion that liver TICs are responsible for tumorigenesis and local recurrence in subgroups of HCC 6, 11. Therefore, targeting such liver TIC-like traits may be a potential strategy to reduce tumor recurrence.

Aberrant activation of Wnt/β-catenin signaling is essential for the self-renewal capacity and drug-resistant properties of human colorectal cancer stem cells, and is associated with poor survival 12, 13. Approximately 30% of cases of HCC exhibit hyperactivation of the Wnt/β-catenin signaling pathway 14-16. It has been reported that aberrant Wnt/β-catenin pathway activation in HCC promotes proliferation and sorafenib resistance 17, whereas Wnt/β-catenin pathway inhibition weakens tumorigenic potential 18, suggesting hyperactivation of the Wnt/β-catenin signaling pathway contributes to tumorigenic properties and the TIC-like phenotype in HCC. Hence, exploration of the molecular mechanisms underlying Wnt/β-catenin signaling hyperactivation in HCC may reveal valuable therapeutic targets to prevent recurrence.

Transcription factor activating enhancer-binding protein 4 (TFAP4), a ubiquitously expressed transcription factor, is upregulated in multiple human malignancies, such as colorectal and gastric carcinoma 19-23, suggesting TFAP4 plays an oncogenic role in cancer pathogenesis and progression. Importantly, Jung et al. demonstrated that overexpressing TFAP4 directly repressed p21 transcription induced by transforming growth factor (TGF)-β in human keratinocytes and reduced p21 expression during myelomonoblast differentiation. They also found that TFAP4 is specifically expressed in colonic progenitor cells and colorectal carcinoma cells and that it can maintain cells in progenitor-like state. These results suggest that TFAP4 may be involved in maintaining cancer cells in a progenitor-like state 24, 25. However, the precise mechanism underlying the oncogenic function of TFAP4 remains largely unknown.

Herein, we report that TFAP4 overexpression promoted Wnt/β-catenin pathway activation by directly binding with the promoters of DVL1 (dishevelled segment polarity protein 1) and LEF1 (lymphoid enhancer binding factor 1) to enhance the tumorigenicity and TIC-like phenotype of HCC cells in vitro and in vivo. These results provide new insight into TFAP4-mediated tumorigenicity and the mechanisms by which the Wnt/β-catenin pathway is hyperactivated in HCC.

Methods

Cell culture

Immortalized LO2 human liver cells were maintained in RPMI-1640 medium (Gibco, Grand Island, NY, USA, Cat. No. 61870044) supplemented with 10% fetal bovine serum (FBS; HyClone, Logan, UT, USA), 100 U/mL penicillin and 100 mg/mL streptomycin in a 5% CO2 atmosphere at 37 °C. HCC cell lines, including HepG2, Hep3B, PLC/PRF-5, MHCC-97H, HCC-LM3, MHCC-97L and QGY-7701 cells, were cultured in DMEM supplemented with 10% FBS.

Patients and tissue specimens

This study was conducted on a total of 197 primary paraffin-embedded, archived HCC specimens that had been histopathologically and clinically diagnosed at the Sun Yat-Sen University Cancer Center from 2003 to 2010. Clinical and clinicopathological classification and staging were determined according to the American Joint Committee on Cancer criteria 26. Detailed clinical information on the 197 patients is summarized in Table . In addition, four freshly non-relapse HCC tissues and four relapse HCC tissues were collected. Percentage tumor purity in sections adjacent to regions used for RNA extraction was estimated during routine histopathologic analysis. Prior Patients' Consents and Approval were obtained from the Institutional Research Ethics Committee (IREC) for the clinical materials used in this study.

Western blotting analysis

Western blotting was performed according to a standard protocol, as described previously 27. The following primary antibodies were used: anti-TFAP4 (Sigma, St. Louis, MO, USA, Cat. No. HP001912), anti-β-catenin (Pharmingen/BD Biosciences, Bedford, MA, USA, Cat. No. 610254), anti-DVL1 (Abcam, Cambridge, MA, USA, Cat. No. ab106844), and anti-LEF1 (Abcam, Cat. No. ab137872). To confirm equal sample loading, the membranes were stripped and re-probed using anti-α-tubulin antibody (Sigma-Aldrich, Cat. No. T6199), anti-β-actin antibody (Abcam, Cat. No. ab6276), or anti-p84 antibody (Abcam, Cat. No. ab487).

Immunohistochemistry

Immunohistochemistry (IHC) was performed in 197 HCC specimens, which were assessed using an anti-TFAP4 antibody (Sigma-Aldrich, Cat. No. HP001912), anti-active- β-catenin antibody (EMD Millipore, Temecula, CA, USA, Cat. No. 05-665-AF555), anti-DVL1 antibody (Proteintech, Rosemont, IL, USA, Cat. No. 27384-1-AP), or anti-LEF1 antibody (Abcam, Cat. No. ab22884) according to a standard protocol as described previously 28, 29. The degree of immunostaining, scored separately by two independent pathologists who were blinded to the histopathological features and patient data of the samples, was determined by the staining index (SI). The SI was calculated as the product of the grade of tumor cell proportions and the staining intensity score. The tumor cell proportions were graded as follows: 0, no positive tumor cells; 1, < 2% positive tumor cells; 2, 2-8% positive tumor cells; 3, 8-20% positive tumor cells; and 4, > 20% positive tumor cells. Staining intensity was scored as: 1, no staining; 2, weak staining (light yellow); 3, moderate staining (yellow-brown); and 4, strong staining (brown). Accordingly, the protein expression as evaluated by SI has a possible score of 0, 1, 2, 3, 4, 6, 8, 9, 12, or 16. Samples with a SI ≥ 8 were determined as high expression, and those with a SI < 8 were determined as low expression. Cutoff values were determined based on a measure of heterogeneity using the log-rank test with respect to overall survival. β-Catenin activation was scored as nuclear positivity if > 5% of cells were stained.

IHC staining for protein expression was also analyzed quantitatively using an Olympus BX51 image analysis system along with Olympus cellSens Dimension 1.5 Imaging software (Olympus, Tokyo, Japan). The stained sections were assessed at 200× magnification, and 10 representative staining fields of view in each section were analyzed to determine mean optical density (MOD), which represents the strength of the staining signals measured as the proportion of positive pixels. MOD values for different groups of tissues were compared using the Student's t-test. Cut-off values were determined based on a measure of heterogeneity using the log-rank test with respect to survival. P < 0.05 was considered significant.

Plasmids, retroviral infection, and transfection

The human TFAP4 gene was PCR-amplified from cDNA and cloned into a pSin-EF2 lentiviral vector. To silence TFAP4, two human TFAP4-targeting short hairpin RNA (shRNA) sequences were cloned into a SUPER.retro.puro vector (OligoEngine, Washington, USA) to generate the respective pSUPER.retro.TFAP4-RNAi(s). HepG2 or LM3 cells were plated at 2×106 cells per p100 plate and transfected with 10 µg of the indicated plasmids. Stable cell lines expressing TFAP4 or TFAP4 shRNA were generated via retroviral infection using HEK293T cells as previously described 30. The stable cell lines were selected for 10 days with 0.5 mg/mL puromycin. The human DVL1 promoter region spanning nucleotides -348 to -193 and the human LEF1 promoter region spanning nucleotides -321 to -173 (relative to the transcription initiation site) generated by PCR-amplification from HepG2 cells were cloned, respectively, into the NheI/BglII sites of pGL3-basic luciferase reporter plasmid (Promega, Madison, WI, USA) to generate DVL1 and LEF1 luciferase reporters. The primers are listed in the Supplemental Methods section on Plasmids, Retroviral Infection and Transfection.

Immunofluorescence assay

Cells were seeded on coverslips and cultured routinely. Once the cells had reached 70% confluence, they were fixed for 15 min in 4% paraformaldehyde (Beyotime, Shanghai, China). After washing with phosphate-buffered saline (PBS) three times, the cells were incubated with 0.1 mL Triton X-100 (0.5%) for 20 min and blocked for 20 min with 10% bovine serum albumin (BSA, Sigma-Aldrich, Cat. No. A1933-1G). Then, the cells were incubated with primary antibodies (anti-β-catenin, Pharmingen/BD Biosciences, Cat. No. 610254) at 4 °C for 8-10 h, washed with PBS, and incubated with fluorescein-conjugated goat anti-rabbit secondary antibodies (GeneCopoeia, Rockville, MD, USA) for 1 h. Nuclei were stained using 4,6-diamidino-2-phenylindole (DAPI, GeneCopoeia). The MOD method was used to quantify the nuclear β-catenin staining signals in the HepG2 cells and LM3 cells. Briefly, the stained slides were evaluated at 400× magnification using the FV10-ASW 4.2 computerized image analysis system (Olympus, Tokyo, Japan) assisted by ImageJ software (Version 1.48, https://imagej.nih.gov/ij/); 10 randomly picked fields per slide were analyzed to determine the nuclear β-catenin MOD, which represents the strength of staining signals as measured per positive pixels. The MOD data were statistically analyzed using the t-test to compare the MOD differences between groups; P < 0.05 was considered statistically significant.

Tumor sphere formation assay

The cells (5×102) were seeded in 6-well ultra-low cluster plates for 10-12 days. The tumor spheres were cultured in DMEM/F12 serum-free medium (Invitrogen, Carlsbad, CA, USA, Cat. No. 88215) supplemented with 2% B27 (Invitrogen, Cat. No. 12587010), 20 ng/mL epidermal growth factor (EGF, PeproTech, Rocky Hill, USA, Cat. No. 37000015), 20 ng/mL basic fibroblast growth factor (bFGF, PeproTech, Cat. No. 100-18B), 5 μg/mL insulin (PeproTech, Cat. No. 100-11), and 0.4% BSA (Sigma-Aldrich, Cat. No. A1933-1G). After 10-12 days, the tumor spheres (tight, spherical, non-adherent masses > 50 µm in diameter) were counted, and their images were captured under an inversed microscope. Sphere formation efficiency was calculated as colonies/input cells×100%.

Flow cytometric analysis

HCC cells were digested with trypsin and resuspended at 1×106 cells per mL in DMEM containing 2% FBS and then pre-incubated at 37 °C for 30 min with or without 100 μM verapamil (Sigma-Aldrich, Cat. No. 1711202) to inhibit ATP-binding cassette (ABC) transporters. The cells were then labeled with 5 μg/mL Hoechst 33342 (Sigma-Aldrich, Cat. No. B2261), and incubated for 90 min at 37 °C and were swirled every 10 min. After 90 min, the cells were washed with ice-cold PBS. After centrifugation using a desk centrifuge (Heraeus, Hanau, Germany), the cells were washed with PBS and processed for flow cytometry analysis. The cells were counterstained with 2 μg/mL propidium iodide (PI; Sigma-Aldrich, Cat. No. P4170) to identify dead cells. Then, 1×106 viable cells were analyzed and sorted in an EPICS ALTRA flow cytometer (Beckman Coulter, Brea, CA, USA). The data were analyzed by Summit5.2 software (Beckman Coulter). All experiments were performed in triplicate. The results were consistent and considered statistically significant for all tests.

Luciferase reporter assay

Luciferase reporter assay was performed according to a standard protocol as described previously 28. Briefly, HCC cells (3×104 cells/well) were seeded in 24-well plates in triplicate and allowed to settle for 24 h. The indicated plasmids and 1.5 ng pRL-TK Renilla plasmid were transfected using Lipofectamine 3000 Reagent (Thermo Fisher Scientific, Waltham, MA, USA, Cat. No. L3000008). At 48 h post-transfection, luciferase and Renilla signals were determined by a Dual Luciferase Reporter Assay Kit (Promega, Cat. No. E1980) according to the manufacturer's instructions as previously described 31.

Preparation of nuclear extracts

Confluent HepG2 and LM3 cells in T75 flasks were washed with 5 mL PBS/phosphatase inhibitors, the supernatant aspirated, and 3 mL ice-cold PBS/phosphatase inhibitors added. The cells were removed by gently scraping with a cell lifter and transferred to a prechilled 15-mL conical tube. The cell suspension was centrifuged for 5 min at 200 rcf in a centrifuge precooled at 4 °C, and the supernatant was discarded. A Nuclear Extract Kit (Active Motif, Rixensart, Belgium, Cat. No. 40010) was used according to the manufacturer's instructions for isolating nuclear extracts from cell pellets.

Chromatin immunoprecipitation-qPCR (ChIP-qPCR)

Chromatin immunoprecipitation (ChIP) was performed as described previously 32. Briefly, crosslinking was performed with 1% formalin, and the cells were lysed in SDS buffer and sonication was used to fragment the DNA. ChIP for TFAP4 was performed using a Flag antibody (Sigma, SAB4301135). Eluted DNA fragments were analyzed by qPCR. The primers are listed in section 'Chromatin immunoprecipitation-qPCR' in Supplementary Material.

Tumor xenograft

Male BALB/c nude mice (5 to 6-weeks-old, 16-18 g) were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China) and were housed in a barrier facility on a 12-h light/dark cycle. Based on a previously described standard protocol 28, the mice were randomly divided into four groups (n = 10 per group). HepG2 cells (1×106, 1×105, 1×104, 1×103, or 1×102 cells), stably transfected with TFAP4-RNAi#1, RNAi-vector, TFAP4 or vector, with Matrigel (final concentration, 25%) were inoculated into the inguinal folds of mice. Tumor volumes were measured with an external caliper, and calculated using the equation (L × W2)/2. At 36 days after inoculation, the mice were sacrificed, and the tumors were excised and subjected to pathologic examination.

Bioinformatics analysis

Gene set enrichment analysis (GSEA, http://software.broadinstitute.org/gsea/msigdb/index.jsp) software programs were used to analyze The Cancer Genome Atlas (TCGA) HCC dataset.

Statistical analysis

All statistical analyses were carried out using SPSS version 21.0 statistical software (SPSS Inc., New York, NY, USA). The two-tailed paired Student's t-test was used to assess the comparisons between groups. The relationships between TFAP4 expression and clinicopathological characteristics were determined using the χ2 test. Survival curves were plotted using the Kaplan-Meier method and compared using the log-rank test. Survival data were evaluated using univariate and multivariate Cox regression analyses. Bivariate correlations between variables were calculated using Spearman's rank correlation coefficients. P-values of < 0.05 were considered statistically significant.

Results

Elevated expression of TFAP4 is associated with recurrence in human HCC

To explore the role of TFAP4 in HCC and the mechanisms underlying the oncogenic function of TFAP4, we initially analyzed TFAP4 mRNA expression between HCC specimens and normal controls in a publicly available HCC dataset (The Cancer Genome Atlas, TCGA). As shown in Figure , TFAP4 mRNA expression was significantly elevated in HCC tissues compared to tumor-adjacent liver tissues. Importantly, TFAP4 mRNA expression was negatively associated with relapse-free survival (P = 0.008, Figure ) and overall survival (P = 0.031, Figure ), suggesting that TFAP4 gene expression might be an indicator of the risk of recurrence in HCC. Consistently, both TFAP4 protein and mRNA expression were significantly higher in the HCC cell lines than in the immortalized LO2 human liver cell line (Figure ) and in the HCC relapse tissues compared with the non-relapse HCC tissues (Figure ).

To evaluate the relationship between TFAP4 expression and the clinicopathological features of HCC, 197 primary HCC tissues (134 primary HCC specimens without relapse and 63 primary HCC specimens that relapsed after treatment) were subjected to IHC staining using a human TFAP4 antibody. TFAP4 expression was significantly higher in the HCC relapse tissues compared to the non-relapse tissues (Figure ). Furthermore, statistical analyses of the 197 specimens revealed that TFAP4 protein expression was associated with clinical stage (P < 0.001), tumor size (P = 0.001), histological differentiation (P = 0.038) and recurrence (P = 0.007), but not with age, gender, or alpha-fetoprotein (AFP), and hepatitis B surface antigen (HBsAg) (Table and Table ). Moreover, Kaplan-Meier analysis and log-rank testing revealed that TFAP4 protein expression was negatively associated with relapse-free survival (P < 0.001; n = 197; Figure ) and overall survival (P < 0.001; Figure ). Moreover, univariate and multivariate analysis revealed that TFAP4 expression was an independent prognostic factor of overall survival of HCC (P < 0.001; Table ). Taken together, these data indicate that upregulation of TFAP4 contributes to recurrence and is associated with a poorer prognosis in HCC.

Overexpression of TFAP4 in HCC cells promotes a TIC-like phenotype in vitro

TFAP4 expression was inversely correlated with tumor differentiation status (Table ), which suggested that TFAP4 may be involved in the regulation of stemness. Consistently, GSEA of a publicly available HCC dataset (TCGA) revealed that TFAP4 expression was positively associated with stemness signatures (Figure ). Furthermore, real-time PCR revealed that the mRNA levels of the stemness-related markers, including CD133, NANOG, SOX2, ABCG2, BMI1, CD44, EpCAM and ALDH1A1 were significantly upregulated in TFAP4-transduced HepG2 and HCC-LM3 cells, and conversely downregulated in TFAP4-silenced HCC cells (Figure ). Next, we conducted the tumor sphere formation assay to examine the effect of TFAP4 on the self-renewal ability of spherogenic HCC cells. Notably, TFAP4-transduced cells formed more spheres with higher cell numbers than the vector control cells (Figure ), whereas TFAP4-silenced cells formed fewer spheres with lower cell numbers than the vector control cells (Figure ). Taken together, our results suggest that TFAP4 overexpression promotes HCC cell tumorigenicity in vitro.

The side population (SP) is a subpopulation of cells that may exhibit a TIC-like phenotype 33. Consistently, overexpression of TFAP4 increased the proportion of HepG2 SP+ cells from 0.99% to 2.39%, and HCC-LM3 SP+ cells from 0.60% to 2.27% (Figure ). Conversely, silencing TFAP4 decreased the proportion of HepG2 SP+ cells from 1.00% to 0.15% and HCC-LM3 SP+ cells from 0.58% to 0.12% (Figure ). Collectively, these results indicate that overexpression of TFAP4 enhances the TIC subpopulation and promotes a TIC-like phenotype in HCC cells in vitro.

Overexpression of TFAP4 enhances the tumorigenicity of HCC cells in vivo

To explore the oncogenic role of TFAP4 in the promotion of the TIC-like phenotype in HCC cells in vivo, we subcutaneously inoculated different numbers of HCC cells mixed with Matrigel into the inguinal folds of BALB/c nude mice. The tumors formed by TFAP4-transduced HepG2 cells were significantly larger than those formed by vector control cells after implantation of 1×106, 1×105, 1×104, 1×103, or 1×102 cells (Figure and Table ). Conversely, TFAP4-silenced cells formed significantly smaller tumors and had lower tumorigenicity (Figure and Table ). Importantly, only TFAP4/HepG2 cells formed tumors when 1×102 cells were implanted (Figure and Table ). These results suggest that TFAP4 overexpression promotes the tumorigenicity of HCC cells in vivo.

Wnt/β-catenin signaling contributes to TFAP4-mediated TIC-like phenotype in HCC cells

To explore the mechanism related to the TFAP4-mediated TIC-like phenotype in HCC, we analyzed the expression of TFAP4 and the genes regulated by various signaling signatures using GSEA of TCGA HCC datasets. TFAP4 mRNA expression correlated positively with Wnt-activated gene signatures (Figure ) and negatively with Wnt-suppressed gene signatures (Figure ), indicating that TFAP4 may activate the Wnt/β-catenin signaling pathway. Furthermore, TOP/FOP flash assays revealed that overexpression of TFAP4 markedly increased whereas TFAP4 knockdown attenuated the transcriptional activation of TCF/LEF in the HepG2 and HCC-LM3 cells (Figure ). Additionally, nuclear extract and immunofluorescence assays showed that TFAP4 overexpression substantially increased the β-catenin nuclear signals, whereas knockdown of TFAP4 reduced them (Figure ). Meanwhile, the mRNA levels of the downstream targets of Wnt/β-catenin signaling, including CCND1, CD44, JUN, and TCF1, were increased in TFAP4-transduced cells but were decreased in TFAP4-silenced cells (Figure ). Collectively, these data suggest that TFAP4 overexpression activates the Wnt/β-catenin signaling pathway.

To further validate the downstream targets of TFAP4 in the Wnt/β-catenin signaling pathway, we performed JASPAR database analysis and predicted that the promoters of the DVL1, LEF1 and TCF4 genes contain TFAP4 binding sites (Figure and Figure ). Chromatin immunoprecipitation (ChIP) assay confirmed that TFAP4 can directly associate with the DVL1 and LEF1 promoters but does not bind to the TCF4 promoter (Figure and Figure ). Furthermore, TFAP4 overexpression significantly increased DVL1/LEF1 promoter-driven reporter activity and DVL1 and LEF1 protein expression levels in HCC cells, whereas silencing TFAP4 had the opposite effects (Figure ), confirming TFAP4 directly targets and positively regulates DVL1 and LEF1. Consistently, overexpressing TFAP4 increased the LEF1 and DVL1 mRNA levels in HCC cells, but silencing TFAP4 decreased them (Figure ). However, no obvious alteration of TCF4 mRNA was observed in TFAP4-dysregulated cells and vector control cells, which was consistent with the ChIP experiment results (Figure ). Next, we investigated whether DVL1 or LEF1 activation was required for the ability of TFAP4 to enhance the TIC-like phenotype in HCC cells. Silencing DVL1 or LEF1 significantly reduced the sphere-forming ability of TFAP4-overexpressing cells and the expression of the stemness-related markers, including CD133, NANOG, SOX2, ABCG2, BMI1, CD44, EpCAM, and ALDH1A1, as well as the expression of the downstream targets of Wnt/β-catenin signaling (Figure and Figure ). The increased proportion of SP+ cells induced by overexpression of TFAP4 was also dramatically abrogated by the DVL1 or LEF1 small interfering RNA (siRNA) (Figure ). These results suggest that activation of DVL1 or LEF1 is required for the ability of TFAP4 to promote a TIC-like phenotype in HCC cells. Taken together, these data indicate that TFAP4 enhances the tumorigenicity of HCC cells by directly targeting and positively regulating DVL1 and LEF1 to activate the Wnt/β-catenin signaling pathway.

TFAP4 expression correlates with Wnt/β-catenin signaling activation in human HCC

Next, we examined whether the Wnt/TFAP4/DVL1-LEF1 axis identified in HCC cells were clinically relevant using IHC analysis. TFAP4 protein expression was correlated positively with the levels of DVL1, LEF1, and activated β-catenin (P < 0.05; n = 87) (Figure ). These observations were validated in seven freshly collected HCC tissue specimens (Figure ), in which TFAP4 expression correlated positively with the expression of DVL1 (P = 0.019, R = 0.698), LEF1 (P = 0.048, R= 0.576), and activated β-catenin (P = 0.021, R = 0.687; Figure ). Taken together, our results indicate that upregulation of TFAP4 in HCC correlates positively with expression DVL1 and LEF1, which in turn activates the Wnt/β-catenin signaling pathway to enhance the tumorigenicity of HCC cells.

Discussion

The key findings of the present report are that TFAP4 promotes tumor sphere formation, increases the proportion of SP+ cells, enhances tumorigenicity and activates the Wnt/β-catenin signaling pathway in HCC. Meanwhile, TFAP4 was overexpressed in human HCC and high TFAP4 expression was associated with poor overall survival and relapse-free survival, indicating TFAP4 may represent a valuable prognostic factor for recurrence in HCC.

TFAP4 belongs to the basic helix-loop-helix-leucine zipper (bHLH-LZ) subgroup of proteins and recognizes the symmetrical DNA core sequence CAGCTG 34, 35. Overexpression of TFAP4 promotes neuroblastoma cell growth 36 and induces the epithelial-mesenchymal transition (EMT) and metastasis in colorectal cancer 32. Importantly, it is believed that TFAP4 is involved in maintaining cancer cells in a progenitor-like state 25, which confers a degree of tumor stemness. TICs, which have self-renewal, multiple differentiation and tumorigenic properties, are critical determinants of the capacity for malignant proliferation, invasion, metastasis, tumor recurrence and drug resistance 37, 38. Several studies have shown that TICs most likely give rise to HCC and a progenitor cell phenotype is detected in a substantial proportion of HCC cases 39, 40. In fact, patients with HCC whose tumors express stem cell markers such as CD133, CD44 and CD90 have a significantly poorer prognosis and higher rate of recurrence 41. We found that overexpressing TFAP4 in HCC cells promoted tumor sphere formation and increased the proportion of SP+ cells (these features are associated with poor overall survival and relapse-free survival in patients with HCC), whereas silencing TFAP4 inhibited sphere formation and reduced the proportion of SP+ cells (which is associated with improved survival). These results suggest that upregulation of TFAP4 enhances tumorigenicity and promotes a TIC-like phenotype in HCC cells, which may hold promise for development of novel therapeutic treatments to prevent recurrence.

The Wnt/β-catenin pathway, also termed the canonical Wnt pathway, is crucial to embryonic development and adult tissue self-renewal 42, 43. Aberrant activation of this pathway can cause uncontrolled cell growth and malignant transformation 42, 43. Malanchi et al. demonstrated that periostin interacts with Wnt1 and Wnt3a, inducing Wnt signaling and sustaining a TIC phenotype in breast tumor cells 44. MMP3 secreted by mammary epithelial cells stimulated Wnt/β-catenin signaling in mammary stem cells 45. Importantly, Zhu et al. found that β-catenin methylation suppressed β-catenin ubiquitination and stability, leading to Wnt/β-catenin signaling activation and sustaining tumorigenicity of HCC cell lines 46. The Wnt/β-catenin pathway also participates in TIC self-renewal and proliferation in HCC cells 47-49. Consistently, we found that Wnt/β-catenin pathway was hyperactivated in human HCC. Herein, we demonstrate that TFAP4 directly targets and positively regulates DVL1 and LEF1, which results in Wnt/β-catenin pathway activation and upregulation of multiple downstream genes. These results reveal a novel molecular mechanism of Wnt/β-catenin pathway hyperactivation in HCC. Therefore, these results suggest that TFAP4 may not only contribute to Wnt/β-catenin activation but may also selectively regulate the downstream targets of β-catenin, which would be worthy of further investigation.

Jung and colleagues reported that the oncoprotein c-MYC could upregulate TFAP4 transcriptionally by directly binding to a region within the first intron of the AP4 gene, which contains four canonical c-MYC binding motifs. They further demonstrated that TFAP4 directly represses p21 by occupying four CAGCTG motifs in the p21 promoter via its basic region. Importantly, TFAP4 overexpression could block p53/p21-induced cell cycle arrest in cells with DNA damage and antagonize p21 induction mediated by TGF-β and 12-O-tetradecanoylphor-bol-13-acetate (TPA) during myelomonoblast differentiation. Therefore, this finding not only describes a novel mechanism by which c-MYC mediates p21 repression but also suggests that TFAP4 may be involved in maintaining cancer cells in a progenitor-like state 24, 25. Moreover, c-MYC-mediated TFAP4 induction is required for c-MYC overexpression-induced cell cycle progression, metastasis, senescence suppression, and transformation, as well as CD8+ T cell activation and sustained expansion of germinal center B cells 50-54. Interestingly, c-MYC is a downstream target of Wnt/β-catenin signaling 55, which is constitutively activated in HCC and in other tumor types 56, 57. Herein, we demonstrate that TFAP4 overexpression activates the Wnt/β-catenin pathway in HCC by directly binding to the DVL1 and LEF1 promoters, thereby enhancing tumorigenicity. Accordingly, these results suggest that TFAP4-mediated c-MYC upregulation might take place be through the activation of Wnt/β-catenin signaling. Therefore, c-MYC and TFAP4 might form a positive feedback loop through Wnt/β-catenin signaling in HCC cells, which would contribute to HCC initiation and progression.

In summary, we demonstrate that TFAP4 overexpression in HCC directly upregulates DVL1 and LEF1 and activates the Wnt/β-catenin pathway, thereby enhancing tumorigenicity, promoting a TIC-like phenotype and conferring a poor prognosis. Further exploration of the biological function of TFAP4 in HCC would increase our knowledge of the mechanisms underlying the high rate of recurrence in HCC, and establish whether TFAP4 represents a prognostic factor or potential therapeutic target for recurrence in HCC.

Table 1

Effect of Transcription factor AP-4 on the tumorigenicity of hepatocellular carcinoma cells in vivo (n = 10/group).

Cell type	Cell number	Tumor incidence	TIC Frequency (95% CI)	P value
HepG2-Vector	1000000	10/10	1/40,070(1/79,901-1/20,095)
	100000	7/10
	10000	5/10
	1000	3/10
	100	0/10		< 0.01
HepG2-TFAP4	1000000	10/10	1/12025(1/26,124-1/5,535)
	100000	9/10
	10000	7/10
	1000	5/10
	100	2/10
HepG2-RNAi-vector	1000000	10/10	1/28,214(1/5,8953-1/13,503)
	100000	8/10
	10000	6/10
	1000	3/10
	100	0/10		< 0.05
HepG2-TFAP4-RNAi#1	1000000	10/10	1/65,094(1/129,823-1/3,2638)
	100000	6/10
	10000	4/10
	1000	1/10
	100	0/10