Forkhead box Q1 promotes hepatocellular carcinoma metastasis by transactivating ZEB2 and VersicanV1 expression

Authors

  • Limin Xia,

    1. State Key Laboratory of Cancer Biology and Xijing Hospital of Digestive Diseases, Fourth Military Medical University, Xi'an Shaanxi Province, P.R. China
    2. Department of Gastroenterology, Tongji Hospital of Tongji Medical College, Huazhong University of Science and Technology, Wuhan Hubei Province, P.R. China
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Wenjie Huang,

    1. State Key Laboratory of Cancer Biology and Xijing Hospital of Digestive Diseases, Fourth Military Medical University, Xi'an Shaanxi Province, P.R. China
    2. Department of Gastroenterology, Tongji Hospital of Tongji Medical College, Huazhong University of Science and Technology, Wuhan Hubei Province, P.R. China
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Dean Tian,

    1. Department of Gastroenterology, Tongji Hospital of Tongji Medical College, Huazhong University of Science and Technology, Wuhan Hubei Province, P.R. China
    Search for more papers by this author
  • Lin Zhang,

    1. State Key Laboratory of Cancer Biology and Xijing Hospital of Digestive Diseases, Fourth Military Medical University, Xi'an Shaanxi Province, P.R. China
    Search for more papers by this author
  • Xingshun Qi,

    1. State Key Laboratory of Cancer Biology and Xijing Hospital of Digestive Diseases, Fourth Military Medical University, Xi'an Shaanxi Province, P.R. China
    Search for more papers by this author
  • Zhangqian Chen,

    1. State Key Laboratory of Cancer Biology and Xijing Hospital of Digestive Diseases, Fourth Military Medical University, Xi'an Shaanxi Province, P.R. China
    Search for more papers by this author
  • Xin Shang,

    1. State Key Laboratory of Cancer Biology and Xijing Hospital of Digestive Diseases, Fourth Military Medical University, Xi'an Shaanxi Province, P.R. China
    Search for more papers by this author
  • Yongzhan Nie,

    1. State Key Laboratory of Cancer Biology and Xijing Hospital of Digestive Diseases, Fourth Military Medical University, Xi'an Shaanxi Province, P.R. China
    Search for more papers by this author
  • Kaichun Wu

    Corresponding author
    1. State Key Laboratory of Cancer Biology and Xijing Hospital of Digestive Diseases, Fourth Military Medical University, Xi'an Shaanxi Province, P.R. China
    • Address reprint request to: Kaichun Wu or Limin Xia, State Key Laboratory of Cancer Biology and Xijing Hospital of Digestive Diseases, Fourth Military Medical University, Xi'an 710032, Shaanxi Province, People's Republic of China. E-mail: wu_kaichun@yahoo.cn or limin_xia@hotmail.com; fax: +86 29 8253 9041.

    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

  • Supported by combined grants from the National Natural Science Foundation of China (No. 81272652, No. 81172290, No. 91129723, No. 81090270, and No. 81090273), the National Key and Basic Research Development Program of China (No. 2010CB529302 and 2010CB529306), the National Municipal Science and Technology Project (2009ZX09103-667 and 2009ZX09301-009-RC06), and the Chinese Postdoctoral Science Foundation (No. 20100471776 and No. 201104757).

Abstract

Forkhead box Q1 (FoxQ1) is a master regulator of tumor metastasis. However, the molecular mechanism of FoxQ1 in regulating hepatocellular carcinoma (HCC) metastasis remains unknown. Here we report a novel function for FoxQ1 in modifying the tumor microenvironment to promote HCC metastasis. FoxQ1 expression was an independent and significant risk factor for the recurrence and survival in two independent cohorts totaling 1,002 HCC patients. FoxQ1 induced epithelial-mesenchymal transition (EMT) through the transactivation of ZEB2 expression by directly binding to the ZEB2 promoter. Knockdown of ZEB2 decreased FoxQ1-enhanced HCC metastasis, whereas up-regulation of ZEB2 rescued the decreased metastasis induced by FoxQ1 knocking down. Additionally, serial deletion, site-directed mutagenesis, and a chromatin immunoprecipitation assays showed that VersicanV1, which promoted HCC metastasis and macrophage attraction, was a direct transcriptional target of FoxQ1. FoxQ1-induced VersicanV1 expression promoted the secretion of chemokine (C-C motif) ligand 2 (CCL2) from HCC cells. Chemotaxis assay showed that the culture media from FoxQ1-overexpressing HCC cells increased the migratory activity of the macrophages. Inhibition of VersicanV1 and CCL2 expression significantly inhibited FoxQ1-mediated macrophage migration. In animal studies, the up-regulation of FoxQ1 in HCC cells promoted HCC metastasis and intratumoral tumor associated macrophage (TAM) infiltration, whereas knockdown of VersicanV1 reduced FoxQ1-mediated HCC metastasis and intratumoral TAM infiltration. Depletion of macrophages using clodronate liposomes dramatically decreased FoxQ1-enhanced HCC metastasis. In human HCC tissues, FoxQ1 expression was positively correlated with ZEB2 and VersicanV1 expression and intratumoral TAM infiltration. Patients with positive coexpression of FoxQ1 and ZEB2, FoxQ1, and VersicanV1, or FoxQ1 and intratumoral TAMs were associated with poorer prognosis. Conclusion: FoxQ1 promotes HCC metastasis by transactivating ZEB2 and VersicanV1 expression, resulting in the induction of EMT and the recruitment of macrophage infiltration. (Hepatology 2014;59:958–973)

Abbreviations
CCL2

chemokine (C-C motif) ligand 2

ChIP

chromatin immunoprecipitation analysis

EMT

epithelial-mesenchymal transition

FoxQ1

forkhead box Q1

HCC

hepatocellular carcinoma

IL-6

interleukin-6

LEF

lymphoid enhancer factor

TAM

tumor-associated macrophage

TCF

T cell factor

TLR2

Toll-like receptor 2

TNF-α

tumor necrosis factor-α

TNM

tumor-node-metastasis

Hepatocellular carcinoma (HCC) is the fifth most common malignancy worldwide and the second leading cause of cancer death in Asia.[1] Although the survival of patients with HCC has improved due to advances in surgical techniques, long-term survival after surgical resection remains low. Metastasis is the major reason for the high mortality of HCC patients after surgical resection.[2] Nonetheless, the molecular mechanisms underlying HCC metastasis remain largely unclear. Thus, it is critical to discover the mechanism of HCC metastasis.

Forkhead box Q1 (FoxQ1, also known as HFH1) is a member of the forkhead transcription factor family.[3] The biological function of FoxQ1 has been implicated in hair follicle morphogenesis and gastric epithelial differentiation.[4] FoxQ1 participates in gastric acid secretion and mucin gene expression.[5] Several recent studies have indicated that increased FoxQ1 expression is correlated with metastasis and poor prognosis for many human cancers, including colon cancer, lung cancer, and breast cancer.[6-9] FoxQ1 provokes an epithelial-mesenchymal transition (EMT) change, gain of stem cell-like properties, and acquisition of resistance to chemotherapy-induced apoptosis.[7] Another study reported that FoxQ1 represses E-cadherin expression by binding to the E-box in its promoter region, and FoxQ1 knockdown blocks transforming growth factor-β (TGF-β)-induced EMT in breast cancer cells.[8] These studies suggest that FoxQ1-mediated EMT plays an important role in promoting cancer metastasis. Accumulating evidence has indicated that EMT plays a prominent role in HCC metastasis.[10] Furthermore, a recent study reported that high FoxQ1 expression was an independent prognostic factor in a cohort of 114 HCC patients.[11] However, whether FoxQ1 contributes to HCC metastasis remains unknown, and the molecular mechanism requires further investigation.

A tumor microenvironment plays a critical role in HCC metastasis. Tumor-associated macrophages (TAMs) are important components of inflammatory infiltrates in HCC.[12] TAMs are derived from circulating monocytic precursors by chemoattractants that are secreted by both tumor cells and stromal cells. TAMs promote tumor metastasis through producing several proinflammatory cytokines, such as tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6).[13] Clinical evidence indicates that the infiltration level of TAMs is associated with poor prognosis in HCC patients.[14] These studies suggest that TAMs infiltration plays an important role in promoting HCC metastasis. Currently, it is unknown whether FoxQ1 can modulate the tumor microenvironment to promote metastasis.

VersicanV1, which is an aggregating chondroitin sulfate proteoglycan, is secreted by both tumor cells and TAMs.[15] VersicanV1 is an important proinflammatory mediator in the tumor microenvironment.[16] Tumor cell secreted-VersicanV1 activates macrophages through Toll-like receptor 2 (TLR2), leading to the production of the metastasis-promoting cytokine TNF-α.[17] A recent study reported that VersicanV1 promotes bladder tumor metastasis through the recruitment of macrophage infiltration in a chemokine (C-C motif) ligand 2 (CCL2)-dependent manner. Inhibition of the CCL2/CCR2 axis dramatically decreased VersicanV1-mediated bladder tumor metastasis and macrophage infiltration.[16] Increased expression of VersicanV1 has been correlated with metastasis and poor survival in many human cancers.[18, 19] Using an EMT polymerase chain reaction (PCR) array in the present study, we found that FoxQ1 up-regulated VersicanV1 expression in HCC cells. Therefore, we hypothesize that FoxQ1-mediated VersicanV1 expression may play a role in inflammation-related HCC metastasis.

Wnt/β-catenin signaling plays a critical role in promoting HCC metastasis.[20] Aberrant activation of the Wnt/β-catenin pathway has been reported in a wide range of HCC patients.[21] The binding of Wnt to cell surface receptors initiates a transduction cascade that stabilizes the transcription coactivator β-catenin, which then enters the nucleus to form a transcriptional complex with T cell factor (TCF) or lymphoid enhancer factor (LEF) to activate the expression of Wnt target genes.[22] Interestingly, recent studies have reported that Wnt/β-catenin signaling transactivates FoxM1 and FoxA2 expression, and the interaction of β-catenin and these Fox proteins promoted tumorigenesis and metastasis.[23, 24] Considering the important roles of both β-catenin and FoxQ1 in tumor metastasis, this finding raised the question of whether FoxQ1 is involved in β-catenin-mediated HCC metastasis.

In this study, we present the first evidence that FoxQ1 promoted HCC metastasis through transactivating ZEB2 and VersicanV1 expression, resulting in the induction of EMT and the recruitment of macrophage infiltration. Furthermore, knockdown of FoxQ1 dramatically decreased β-catenin-enhanced HCC metastasis.

Materials and Methods

Plasmid Construction

Plasmid construction was performed according to standard procedures as outlined in our previous study.[25] The primers are presented in Supporting Table S10. For example, the ZEB2 promoter construct (−1247/+109)ZEB2 was generated from human genomic DNA. This construct corresponds to the sequence from −1247 to +105 (relative to the transcriptional start site) of the 5′-flanking region of the human ZEB2 gene. It was generated with forward and reverse primers incorporating SacI and XhoI sites at the 5′ and 3′-ends, respectively. The PCR product was cloned into the SacI and XhoI sites of the pGL3-Basic vector (Promega). The 5′-flanking deletion constructs of the ZEB2 promoter, ((−1106/+109)ZEB2, (−1054/+109)ZEB2, (−853/+109)ZEB2, and (−551/+109)ZEB2), were similarly generated using the (−1247/+109)ZEB2 construct as the template. The FoxQ1 binding sites in the ZEB2 promoter were mutated using the QuikChange II Site-Directed Mutagenesis Kit (Stratagene). The constructs were confirmed by DNA sequencing. Other promoter constructs were cloned in the same manner.

Construction of Lentivirus and Stable Cell Lines

Based on the FoxQ1 sequence (NM_033260.3), three short hairpin RNAs (shRNAs) were designed using the siRNA Target Finder (InvivoGen): shFoxQ1-1, 5′-CTACTCGTACATCGCGCTC AT-3′; shFoxQ1-2, 5′-CCAGCTCCTTCGCCATCG ACA-3′; and shFoxQ1-3, 5′-CGAGTACCTCATGGG CAAGTT-3′. Lentiviral vectors encoding shRNAs were generated using PLKO.1-TRC (Addgene) and designated LV-shFoxQ1-1, LV-shFoxQ1-2, LV-shFoxQ1-3, and LV-shcontrol. Lentiviral vectors encoding the human FoxQ1 gene were constructed in FUW-teto (Addgene) and designated LV-FoxQ1. An empty vector was used as the negative control and was designated LV-control. The lentiviral vectors were transfected into the HCC cells with a multiplicity of infection (MOI) ranging from 30 to 50 in the presence of polybrene (6 μg/mL). At 48 hours after transfection, 2.5 μg/mL puromycin (OriGene) was added and the cells were incubated for 2 weeks to select for transfected cells.

A detailed description of the Materials and Methods used in this study can be found in the online Supporting Material.

Results

Overexpression of FoxQ1 Indicates Poor Prognosis and Promotes HCC Metastasis

To explore the potential role of FoxQ1 in determining the clinical outcomes of HCC patients, we assessed its expression in a tissue microarray of 690 HCC patients (Cohort I). The immunohistochemistry results showed that FoxQ1 was primarily localized to the nucleus (Fig. 1A1). Positive FoxQ1 expression was found in 376 of 690 (54.49%) primary HCC samples, compared with only 41 of 690 (5.94%) adjacent nontumor tissues (P < 0.001). FoxQ1 overexpression was significantly correlated with multiple tumor numbers, microvascular invasion, poor tumor differentiation, and a higher tumor-nodule-metastasis (TNM) stage (Supporting Table S1). Patients with positive FoxQ1 expression had shorter overall survival and higher recurrence rates than patients with negative FoxQ1 expression (Fig. 1B1). Multivariate analysis indicated that FoxQ1 expression was an independent and significant factor for recurrence and reduced survival (Supporting Table S2). The prognostic value of FoxQ1 protein expression was validated in an independent cohort of 312 HCC patients (Cohort II) by immunohistochemical staining. Similarly, FoxQ1 expression was significantly correlated with poor prognosis (Fig. 1B2). FoxQ1 overexpression was correlated with microvascular invasion, absent tumor encapsulation, and TNM stage (Supporting Table S1). Multivariate analysis identified FoxQ1 expression as an independent predictor for postoperative recurrence and overall survival (Supporting Table S3).

Figure 1.

Overexpression of FoxQ1 indicates poor prognosis and promotes HCC metastasis. (A1) Representative FoxQ1 expression in adjacent nontumorous tissues and primary and metastatic HCC tissues detected by immunohistochemical methods. (A2) Real-time PCR analysis of FoxQ1 expression in normal liver (n = 10); 90 pairs of HCCs and adjacent nontumorous tissues. (A3) Relative mRNA expression of FoxQ1 in HCC patient samples with recurrence (n = 51) or without recurrence (n = 39). (A4) Relative mRNA expression of FoxQ1 in HCC patient samples with metastasis (n = 40) or without metastasis (n = 40). (B1) Kaplan-Meier analysis of the correlation between FoxQ1 expression and the recurrence or overall survival of 690 HCC patients (Cohort I). (B2) Kaplan-Meier analysis of the correlation between FoxQ1 expression and the recurrence or overall survival of 312 HCC patients (Cohort II). *P < 0.05. (C1) Real-time PCR and western blot analysis of FoxQ1 expression in different HCC cell lines. (C2) After HCC cells were infected with LV-FoxQ1 or LV-shFoxQ1-1, −2, or −3, the level of FoxQ1 protein expression was detected by western blot analysis. (D) Up-regulation of FoxQ1 expression enhanced SMMC7721 (low metastatic potential) cell migration and invasion, whereas down-regulation of FoxQ1 expression decreased HCCLM3 (high metastatic potential) cell migration and invasion. (E) In vivo metastasis assays. Four stable cell lines were transplanted into the livers of nude mice. (E1) Representative bioluminescent imaging (BLI) of the different groups is shown at 10 weeks following orthotopic implantation. (E2) The incidences of lung metastases in the different groups of nude mice. (E3) The overall survival times of the nude mice in the different groups. (E4) The numbers of lung metastatic foci in each group. (E5) Representative H&E staining of lung tissues from the different groups is shown. *P < 0.05.

In a cohort of 90 paired HCC specimens, FoxQ1 messenger RNA (mRNA) was up-regulated in HCC tissues compared with adjacent nontumorous tissues (Fig. 1A2). Patients with recurrences of HCC (51 of 90) exhibited higher FoxQ1 mRNA expression than patients without recurrences (39 of 90) (Fig. 1A3). In addition, FoxQ1 mRNA expression was much higher in primary HCC tissues from patients who developed metastases than those who did not develop metastasis (Fig. 1A4). These studies suggest that FoxQ1 may contribute to HCC progression and metastasis.

The FoxQ1 expression in highly metastatic HCC cell lines was significantly higher than in HCC cell lines with low metastatic potential (Fig. 1C1). Up-regulation of FoxQ1 expression enhanced the migration and invasion of the SMMC7721 cells. Conversely, the inhibition of FoxQ1 expression in HCCLM3 cells decreased cell migration and invasion (Fig. 1D). The in vivo metastatic assay showed that the up-regulation of FoxQ1 increased the incidence of lung metastasis and the number of metastatic lung nodules while decreasing the overall survival time of the SMMC7721-FoxQ1 group. In contrast, the down-regulation of FoxQ1 decreased the incidence of lung metastasis and the number of metastatic lung nodules while increasing the overall survival time of the HCCLM3-shFoxQ1 group (Fig. 2E). In addition, representative hematoxylin and eosin (H&E) staining showed that intrahepatic metastatic nodules were observed in the liver of SMMC7721-FoxQ1 group, whereas no obvious metastatic nodules were observed in the adjacent tumor tissues of the SMMC7721-control group. In contrast, there are intrahepatic metastatic nodules in the liver of the HCCLM3-shcontrol group, whereas no metastatic nodules were found in the adjacent tumor tissues of HCCLM3-shFoxQ1 group (Supporting Fig. S4). Bioluminescent imaging of dissected organs showed that orthotopic SMMC7721-FoxQ1 and HCCLM3-shcontrol cells spread to the peritoneal cavity and invaded intestine organs (Supporting Fig. S5A). Histological examination showed that orthotopic SMMC7721-FoxQ1 and HCCLM3-shcontrol cells invaded intestine tissues (Supporting Fig. S5B). These studies indicate that FoxQ1 promotes HCC invasion and metastasis.

Figure 2.

FoxQ1 promotes HCC metastasis through transactivating ZEB2 expression. (A1) Immunofluorescence staining of an epithelial marker (E-cadherin) and a mesenchymal marker (vimentin) in SMMC7721-FoxQ1 and HCCLM3-shFoxQ1 cells. (A2) Western blot analysis of epithelial and mesenchymal markers in SMMC7721-FoxQ1 and HCCLM3-shFoxQ1 cells. (B) ZEB2 is essential for the FoxQ1-induced reduction of E-cadherin expression. (B1) Western blot and (B2) real-time PCR were used to detect the expression of FoxQ1, ZEB2, and E-cadherin. The knockdown of ZEB2 expression attenuated the loss of E-cadherin expression induced by FoxQ1, whereas the up-regulation of ZEB2 significantly inhibited the increase in E-cadherin expression in HCCLM3-shFoxQ1 cells. (C1) FoxQ1 transactivates ZEB2 promoter activities. The ZEB2 promoter luciferase construct (−1247/+109)ZEB2 was cotransfected with pCMV-FoxQ1, and promoter activities were detected using a luciferase reporter assay. (C2) FoxQ1 inhibits E-cadherin transcription through the up-regulation of ZEB2 expression. Cells were pretransfected with ZEB2 siRNA or control siRNA. After 48 hours, the E-cadherin promoter luciferase construct (pGL3-E-cadherin) was cotransfected with pCMV-FoxQ1. A luciferase reporter assay was used to detect promoter activities. (C3) Deletion and selective mutation analyses identified FoxQ1-responsive regions in the ZEB2 promoter. Serially truncated and mutated ZEB2 promoter constructs were cotransfected with pCMV-FoxQ1 and relative luciferase activities were determined. (C4) A ChIP assay demonstrated the direct binding of FoxQ1 to the ZEB2 promoter in HCC cells. (C5) FoxQ1 binds to the ZEB2 promoter in HCC tissues. The hepatocytes were separated from the liver tissues of HCC patients and healthy controls (HC). Real-time PCR was performed to detect the amounts of immunoprecipitated products. (D) Following the infection of the SMMC7721-FoxQ1 cells and HCCLM3-shFoxQ1 cells with the lentivirus LV-shZEB2 or LV-ZEB2, respectively, the cell migration and invasion capacities were assessed using transwell assays. (E) In vivo metastatic assay. (E1) Cell lines were transplanted into the livers of nude mice. Representative bioluminescent imaging (BLI) of the different groups is shown at 10 weeks following orthotopic implantation. (E2) The incidence of lung metastases in the different groups of nude mice. (E3) The overall survival of the nude mice in the different groups. (E4) The number of lung metastatic foci in each group. (E5) Representative H&E staining of the lungs from the different groups is shown. *P < 0.05.

FoxQ1 Induces EMT Through the Transactivation of ZEB2 Expression

Immunofluorescence showed that the up-regulation of FoxQ1 decreased E-cadherin expression and increased vimentin expression in SMMC7721 cells, whereas the down-regulation of FoxQ1 increased E-cadherin expression and decreased vimentin expression in HCCLM3 cells (Fig. 2A1). Western blot analysis showed that the up-regulation of FoxQ1 in SMMC7721 cells resulted in the decreased expression of epithelial markers (E-cadherin and ZO-1) and the increased expression of mesenchymal markers (vimentin and fibronectin). In contrast, the down-regulation of FoxQ1 in HCCLM3 cells significantly increased the expression of epithelial markers and decreased the expression of mesenchymal markers (Fig. 2A2). These results suggest that FoxQ1 induces EMT in HCC cells.

To obtain insight into the mechanisms through which FoxQ1 affects HCC metastasis, the mRNA expression profiles of SMMC7721-FoxQ1 cells were compared with the profiles of SMMC7721-control cells using a human EMT RT[2] Profiler PCR Array containing 87 EMT-related genes. FoxQ1 overexpression increased the expression of a number of EMT-related genes, including VCAN, ZEB2, TGFB1, IGFBP4, VIM, MMP2, ITGAV, CTNNB1, SNAI1, BMP1, and ITGA5 (Supporting Table S4). Of particular interest were ZEB2 and VCAN, which were up-regulated 4.38- and 5.12-fold.

Considering the critical role of ZEB2 in metastasis, we determined whether ZEB2 was involved in FoxQ1-mediated HCC metastasis. FoxQ1 up-regulated ZEB2 expression and decreased E-cadherin expression in SMMC7721 cells, whereas the inhibition of ZEB2 expression attenuated the loss of E-cadherin expression induced by FoxQ1. In contrast, the knockdown of FoxQ1 decreased ZEB2 expression and increased E-cadherin expression in HCCLM3 cells, whereas the up-regulation of ZEB2 dramatically inhibited the increase in E-cadherin expression in HCCLM3-shFoxQ1 cells (Fig. 2B1,B2). FoxQ1 transactivated ZEB2 promoter activity but inhibited E-cadherin transcription (Fig. 2C1), whereas knockdown of ZEB2 partially relieved the suppression of E-cadherin promoter transcription induced by FoxQ1 (Fig. 2C2). These studies suggest that FoxQ1 inhibits E-cadherin expression by up-regulating ZEB2 expression.

The FoxQ1 consensus binding sequence is 5′-T(A/C)AA(C/T)A-3′.[5] Four putative FoxQ1 binding sites were found in the ZEB2 promoter. To determine the roles of the cis-regulatory elements of the ZEB2 promoter in response to FoxQ1 regulation, a series of ZEB2 promoter truncation mutants were generated. A deletion from nt-1054 to nt-551 significantly decreased the FoxQ1-induced ZEB2 promoter activity (Fig. 2C3), indicating that the sequence that is located between nt-1054 and −551 is critical for the activation of the ZEB2 promoter. Two FoxQ1-binding sites are located in this region. The mutation of both of the FoxQ1-binding sites significantly reduced the FoxQ1-induced transactivation of the ZEB2 promoter (Fig. 2C3). A chromatin immunoprecipitation (ChIP) assay confirmed the direct binding of FoxQ1 to the ZEB2 promoter in HCC cells (Fig. 2C4) and human HCC tissues (Fig. 2C5). These studies suggest that FoxQ1 induces EMT by transactivating ZEB2 expression in HCC cells.

ZEB2 Is Essential for FoxQ1-Mediated HCC Metastasis

The down-regulation of ZEB2 significantly reduced FoxQ1-enhanced cell migration and invasion, whereas the up-regulation of ZEB2 rescued the decreased migration and invasion abilities induced by FoxQ1 knockdown (Fig. 2D). The in vivo metastatic assay showed that the down-regulation of ZEB2 decreased the incidence of lung metastasis and the number of metastatic lung nodules while increasing the overall survival time of the SMMC7721-FoxQ1 group. In contrast, the up-regulation of ZEB2 rescued the decreased incidence of lung metastasis and the number of metastatic lung nodules while decreasing the overall survival time of the HCCLM3-shFoxQ1 group (Fig. 2E). Thus, ZEB2 is essential for FoxQ1-mediated HCC metastasis.

We further evaluated the possible association between FoxQ1 and ZEB2 or E-cadherin expression in human HCC tissues (Cohort I). Both the overexpression of ZEB2 and the down-regulation of E-cadherin were associated with poor prognosis (Fig. 3C1,C2) and aggressive tumor behavior (Supporting Tables S5, S6). Immunohistochemistry revealed that FoxQ1 expression was positively correlated with ZEB2 expression but inversely correlated with E-cadherin expression (Fig. 3A,B). Patients with positive coexpression of FoxQ1 and ZEB2 had the highest recurrence rates and shortest overall survival times (Fig. 3C3), whereas patients with the FoxQ1(+)/E-cadherin(−) expression pattern had the highest recurrence and shortest overall survival (Fig. 3C4).

Figure 3.

FoxQ1 is positively correlated with ZEB2 expression but inversely correlated with E-cadherin expression in human HCC tissues. (A) An immunohistochemical analysis of FoxQ1, ZEB2, and E-cadherin expression in HCC tissues and adjacent nontumorous tissues. (B) The association between the expression of FoxQ1 and either ZEB2 or E-cadherin in Cohort I HCC patients. (C1,C2) A Kaplan-Meier analysis of ZEB2 or E-cadherin expression in Cohort I (n = 690) HCC patients after curative resection. (C3) The Kaplan-Meier analysis of concurrent FoxQ1 and ZEB2 expression with recurrence and overall survival in Cohort I. (C4) The correlation of FoxQ1/E-cadherin coexpression with recurrence and overall survival in Cohort I. (D) The association between the expression of FoxQ1 and either ZEB2 or E-cadherin in Cohort II (n = 312) HCC patients. (E1,E2) A Kaplan-Meier analysis of ZEB2 or E-cadherin expression in Cohort II HCC patients. (E3) The correlation of FoxQ1/ZEB2 coexpression with recurrence and overall survival in Cohort II. (E4) The correlation of FoxQ1/E-cadherin coexpression with recurrence and overall survival in Cohort II.

The expression correlation and prognostic value of FoxQ1, ZEB2, and E-cadherin were validated in an independent cohort of 312 HCC patients (Cohort II) by immunohistochemical staining. Similarly, FoxQ1 was positively correlated with ZEB2 but inversely correlated with E-cadherin expression (Fig. 3D). Both the FoxQ1(+)/ZEB2(+) and FoxQ1(+)/E-cadherin(−) expression patterns were associated with poorer prognosis (Fig. 3E3,E4). Taken together, both the experimental and clinical evidence suggest that the FoxQ1-mediated ZEB2/E-cadherin signaling pathway promotes HCC metastasis and indicates poor prognosis.

VersicanV1 Is a Direct Transcriptional Target of FoxQ1

We questioned whether FoxQ1 regulates VCAN transcription in HCC cells. FoxQ1 up-regulated VersicanV1 expression in SMMC7721 cells, whereas the knockdown of FoxQ1 decreased VersicanV1 expression in HCCLM3 cells (Fig. 4A1). To determine whether FoxQ1 regulates VCAN transcription, a VCAN promoter luciferase construct (−1468/+56)VCAN was cotransfected with pCMV-FoxQ1. A luciferase reporter assay showed that FoxQ1 transactivated VCAN promoter activity (Fig. 4A2). Sequence analysis revealed four putative FoxQ1 binding sites in the VCAN promoter. Serial deletion and site-directed mutagenesis showed that the third and fourth FoxQ1 binding sites were critical for FoxQ1-induced VCAN transactivation (Fig. 4A3). A ChIP assay further confirmed that FoxQ1 binds directly to the VCAN promoter in HCC cells (Fig. 4A4) and human HCC tissues (Fig. 4A5). These studies suggest that VersicanV1 is a direct transcriptional target of FoxQ1.

Figure 4.

VersicanV1, a direct transcriptional target of FoxQ1, promotes HCC metastasis. (A1) After SMMC7721 and HCCLM3 cells were infected with LV-FoxQ1 or LV-shFoxQ1, the mRNA and protein levels of VersicanV1 were detected using real-time PCR and western blot techniques, respectively. (A2) FoxQ1 transactivates the VCAN promoter. The VCAN promoter construct was cotransfected with pCMV-FoxQ1, and the relative luciferase activity was determined. (A3) Deletion and selective mutation analysis identified two FoxQ1-responsive regions in the VCAN promoter. Serially truncated and mutated VCAN promoter constructs were cotransfected with pCMV-FoxQ1, and the relative luciferase activity was determined. (A4-A5) A ChIP assay demonstrated the direct binding of FoxQ1 to the VCAN promoter in (A4) HCC cells and (A5) HCC tissues. (B1) The expression of VersicanV1 in different HCC cell lines was detected by western blot. (B2) After SMMC7721 and HCCLM3 cells were infected with LV-VersicanV1 and LV-shVersicanV1, respectively, the expression of VersicanV1 was detected by western blot. (C) The migration and invasion abilities of the indicated cells were detected by transwell assay. (D) In vivo metastasis assays. The above 4 cell lines were transplanted into the livers of nude mice. (D1) Representative bioluminescent imaging (BLI) of the different groups is shown at 10 weeks following orthotopic implantation. (D2) The incidence of lung metastases in the different groups of nude mice. (D3) The overall survival of the nude mice in the different groups. (D4) The number of lung metastatic foci in each group. (D5) Representative H&E staining of lung tissues from the different groups is shown. (E1-E2) VersicanV1 expression in HCC cells promotes macrophage attraction in a CCL2-dependent manner. (E1) BMDMs or (E2) U937 cells were subjected to migration and transendothelial migration assays toward conditioned media from indicated HCC cells with anti-CCL2 (5 μg/mL) or rCCL2 (10 ng/mL). Attracted cells on the undersurface of the filters were counted. Bars represent the means ± SEM of three independent experiments. (F1-F2) Macrophage infiltration in the primary HCC nodules of different groups was determined by F4/80 immunostaining. (F1) Representative immunohistochemical images of different groups are shown. (F2) Bars represent the means ± SEM of macrophage number, counted in six random high-power fields. (F3) The mRNA levels of human and murine CCL2, TNF-α, IL-6, and IL-8 (MIP2, the murine homolog of human IL-8) in the primary HCC nodules of different groups were detected by real-time PCR. Bars represent the means ± SEM of samples (n = 10) of the indicated groups. *P < 0.05.

VersicanV1 Expression in HCC Cells Promotes the Recruitment of Macrophages and HCC Metastasis

VersicanV1 expression was much higher in HCC cell lines with high metastatic potential than in HCC cell lines with low metastatic potential (Fig. 4B1). To determine whether VersicanV1 regulates the invasion and metastasis of HCC cells, SMMC7721 and HCCLM3 cells were infected with lentivirus LV-VeriscanV1 and LV-shVersicanV1 (Fig. 4B2). The up-regulation of VersicanV1 significantly increased the migration and invasion of SMMC7721 cells, whereas the down-regulation of VersicanV1 decreased the migration and invasion abilities of HCCLM3 cells (Fig. 4C). The in vivo metastatic assay showed that the up-regulation of VersicanV1 increased the incidence of lung metastasis and the number of metastatic lung nodules while decreasing the overall survival time of the SMMC7721-VersicanV1 group. In contrast, the down-regulation of VersicanV1 decreased the incidence of lung metastasis and the number of metastatic lung nodules while increasing the overall survival time of the HCCLM3-shVersicanV1 group (Fig. 4D1-D5). These studies suggest that VersicanV1 promotes HCC invasion and metastasis.

Recent studies reported that VersicanV1 promotes breast cancer and bladder cancer metastasis through the recruitment and activation of macrophage.[16, 17] To determine the chemotactic properties of HCC cell secreted-VersicanV1 on macrophages, we examined the effect of conditioned medium from SMMC7721-VersicanV1 and HCCLM3-shVersicanV1 cells on the migration and transendothelial migration of bone marrow-derived macrophage (BMDMs) or macrophage cell line U937. Conditioned medium from SMMC7721-VersicanV1 cells significantly increased the migration and transendothelial migration abilities of BMDMs and U937 cells compared with the medium from SMMC7721-control cells (Fig. 4E1, E2). CCL2 is a major chemoattractant for macrophages.[26] Enzyme-linked immunosorbent assay (ELISA) analysis confirmed that the up-regulation of VersicanV1 increased CCL2 secretion from SMMC7721 cells, whereas the down-regulation of VersicanV1 decreased CCL2 secretion from HCCLM3 cells (Supporting Fig. S1A). A neutralizing antibody against CCL2 significantly decreased the chemotactic potency of medium from SMMC7721-VersicanV1 cells (Fig. 4E1,E2). In addition, conditioned medium from HCCLM3-shVersicanV1 cells decreased the migration and transendothelial migration abilities of BMDMs and U937 cells compared with the medium from HCCLM3-shcontrol cells, whereas recombinant CCL2 (rCCL2) rescued the decreased chemotactic potency of the medium from HCCLM3-shVersicanV1 cells (Fig. 4E1,E2). These studies suggest that VersicanV1 expression in HCC cells promotes macrophage attraction in a CCL2-dependent manner.

Importantly, we measured macrophage content in transplanted HCC tissues in nude mice by staining for the mature macrophage marker F4/80. The transplantation of SMMC7721 cells that overexpress VersicanV1 (SMMC7721-VersicanV1) significantly increased macrophage infiltration. However, the transplantation of HCCLM3 cells that down-regulate VersicanV1 (HCCLM3-shVersicanV1) decreased macrophage infiltration (Fig. 4F1,F2). Real-time PCR showed that the mRNA levels of CCL2, TNF-α, IL-6, and IL-8 (MIP2, the murine homolog of human IL-8,[27]) all of which are implicated in macrophage recruitment and inflammation, were significantly up-regulated in VersicanV1-overexpressing (SMMC7721-VersicanV1) HCC tissues compared with control HCC tissues (SMMC7721-control), whereas the mRNA levels of CCL2, TNF-α, IL-6, and IL-8 were dramatically decreased in VersicanV1-down-regulating (HCCLM3-shVersicanV1) HCC tissues as compared with control HCC tissues (HCCLM3-shcontrol) (Fig. 4F3).

FoxQ1 Expression in HCC Cells Promotes Macrophage Infiltration and HCC Metastasis by Up-Regulating VersicanV1

To test whether VersicanV1 is involved in FoxQ1-mediated metastasis, we stably down-regulated VersicanV1 in SMMC7721-FoxQ1 cells and up-regulated VersicanV1 in HCCLM3-shFoxQ1 cells (Fig. 5A). The down-regulation of VersicanV1 significantly decreased FoxQ1-enhanced migration and invasion abilities, whereas the up-regulation of VersicanV1 rescued the decreased migration and invasion abilities induced by knocking down FoxQ1 (Fig. 5B). The in vivo metastatic assay showed that the down-regulation of VersicanV1 decreased the incidence of lung metastasis and the number of metastatic lung nodules while increasing the overall survival time of the SMMC7721-FoxQ1 group. In contrast, the up-regulation of VersicanV1 rescued the decreased incidence of lung metastasis and the number of metastatic lung nodules while decreasing the overall survival time of the HCCLM3-shFoxQ1 group (Fig. 5C1-C5). These studies suggest that FoxQ1 promotes HCC metastasis through up-regulating VersicanV1 expression.

Figure 5.

FoxQ1 promotes HCC metastasis through the up-regulation of Versican V1 expression. (A,B) Following the infection of the SMMC7721-FoxQ1 cells and HCCLM3-shFoxQ1 cells with the lentivirus LV-shVersicanV1 or LV-VersicanV1, respectively, (A) the protein levels of FoxQ1 and VersicanV1 were detected by western blot, and (B) the cell migration and invasion capacities were assessed using transwell assays. (C) In vivo metastatic assay. (C1) Cell lines were transplanted into the livers of nude mice. Representative bioluminescent imaging (BLI) of the different groups is shown at 10 weeks following orthotopic implantation. (C2) The incidence of lung metastases in the different groups of nude mice. (C3) The overall survival of the nude mice in the different groups. (C4) The number of lung metastatic foci in each group. (C5) Representative H&E staining of the lungs from the different groups is shown. (D1,D2) FoxQ1 expression in HCC cells promotes macrophage attraction through the VersicanV1/CCL2 axis. (D1) BMDMs were subjected to migration and transendothelial migration assays toward conditioned medium from SMMC7721-FoxQ1 cells treated with LV-shVersicanV1 or anti-CCL2 (5 μg/mL). (D2) BMDMs were subjected to migration and transendothelial migration assays toward conditioned medium from HCCLM3-shFoxQ1 cells treated with LV-VersicanV1 or rCCL2 (10 ng/mL). Attracted cells on the undersurface of the filters were counted. Bars represent the means ± SEM of three independent experiments. (E1,E2) Macrophage infiltration in primary HCC nodules of different groups was determined by F4/80 immunostaining. (E1) Representative immunohistochemical images of different groups are shown. (E2) Bars represent the means ± SEM of macrophage number, counted in six random high-power fields. (E3) Real-time PCR analysis of the mRNA levels of murine and human CCL2, TNF-α, IL-6, and IL-8 in primary HCC nodules from different groups. Bars represent the means ± SEM of samples (n = 10) from the indicated groups. *P < 0.05.

Next, we determined whether FoxQ1 expression in HCC cells promotes macrophage attraction. Conditioned medium from SMMC7721-FoxQ1 cells significantly increased the migration and transendothelial migration abilities of BMDMs compared with the medium from SMMC7721-control cells. ELISA analysis confirmed that the down-regulation of VersicanV1 decreased the secretion of CCL2 from SMMC7721-FoxQ1 cells, whereas the up-regulation of VersicanV1 rescued the decreased secretion of CCL2 from HCCLM3-shFoxQ1 cells (Supporting Fig. S1B). Both the down-regulation of VersicanV1 and the neutralizing antibody against CCL2 significantly decreased the chemotactic potency of medium from SMMC7721-FoxQ1 cells (Fig. 5D1). In contrast, conditioned medium from HCCLM3-shFoxQ1 cells decreased the migration and transendothelial migration abilities of BMDMs compared with the medium from HCCLM3-shcontrol cells, whereas the up-regulation of VersicanV1 and recombinant CCL2 (rCCL2) rescued the decreased chemotactic potency of medium from HCCLM3-shFoxQ1 cells (Fig. 5D2). Similar results were also found in U937 cells (Supporting Fig. S2).

In addition, the overexpression of FoxQ1 in HCC cells significantly increased macrophage infiltration and the expression of inflammatory mediators (CCL2, TNF-α, IL-6, and IL-8), whereas the down-regulation of VersicanV1 decreased the increased macrophage infiltration and inflammatory mediators induced by FoxQ1 overexpression. In contrast, the knockdown of FoxQ1 in HCC cells decreased the macrophage infiltration and inflammatory mediators, whereas the up-regulation of VersicanV1 rescued the decreased macrophage infiltration and inflammatory mediators induced by knocking down FoxQ1 (Fig. 5E1-E3). These studies suggest that FoxQ1 expression in HCC cells promotes macrophage infiltration through the VersicanV1/CCL2 axis.

Depletion of Macrophages Decreases FoxQ1-Mediated HCC Metastasis

To investigate whether macrophage recruitment contributes to FoxQ1-mediated metastasis, we depleted macrophages using clodronate liposomes.[28] Clodronate liposome treatment effectively reduced the presence of F4/80-positive macrophages to nearly undetectable levels in the transplanted SMMC7721-FoxQ1 tumors compared with control liposome treatment (Fig. 6E). Macrophage depletion by clodronate liposome treatment decreased the incidence of lung metastasis and the number of metastatic lung nodules while increasing the overall survival time of the SMMC7721-FoxQ1 group (Fig. 6A-D). The expressions of host and tumor CCL2, TNF-α, IL-6, and IL-8 in the transplanted SMMC7721-FoxQ1 tumors were significantly decreased after clodronate treatment (Fig. 6F). These studies suggest that macrophage infiltration is essential for FoxQ1-mediated HCC metastasis.

Figure 6.

Macrophage depletion inhibits FoxQ1-mediated HCC metastasis. SMMC7721-FoxQ1 cells were implanted into the left lobes of the livers of the nude mice, which had either been previously depleted of macrophages by clodronate-encapsulated liposomes (n = 10) or treated with control liposomes (n = 10). (A) Representative bioluminescent imaging (BLI) of the different groups is shown at 10 weeks following orthotopic implantation. (B) The overall survival of the nude mice in the different groups. (C) The number of lung metastatic foci in each group. (D) Representative H&E staining of lung tissues from the different groups is shown. (E) Immunohistochemical images of transplanted tumor tissue sections stained for mature macrophage marker F4/80. Bars represent the means ± SEM of macrophage number, counted in six random high-power fields. (F) The mRNA levels of human and murine CCL2, TNF-α, IL-6, and IL-8 in primary HCC nodules from different groups were detected by real-time PCR. Bars represent the means ± SEM of samples (n = 10) from the indicated groups. *P < 0.05.

Immunohistochemical results from both independent cohorts showed that VersicanV1 expression was significantly up-regulated in HCC tissues, as compared with adjacent nontumorous tissues, and that VersicanV1 was mainly localized in the cytoplasm (Fig. 7A). Both VersicanV1 overexpression and intratumorous TAMs infiltration were significantly correlated with poorer tumor differentiation and higher TNM stage (Supporting Tables S7, S8). HCC patients with positive expression of VersicanV1 had shorter overall survival times and higher recurrence rates than patients with negative expression of VersicanV1 (Fig. 7C1,E1). Similarly, intratumoral TAMs infiltration was associated with poor prognosis (Fig. 7C2,E2). In addition, FoxQ1 expression was positively correlated with both VersicanV1 expression and intratumoral TAMs infiltration (Fig. 7B,D). Patients with positive coexpression of FoxQ1 and VersicanV1 had the highest recurrence rates and the lowest overall survival times (Fig. 7C3,E3). Similarly, patients with positive coexpression of FoxQ1 and intratumoral TAMs infiltration were associated with poorer prognosis (Fig. 7C4,E4).These studies suggest that FoxQ1-mediated VersicanV1 overexpression and macrophage infiltration promote HCC metastasis and indicate poor prognosis.

Figure 7.

FoxQ1 is positively correlated with both VersicanV1 expression and macrophage infiltration in human HCC tissues. (A) Representative immunohistochemical images of FoxQ1, VersicanV1, and macrophage infiltration in HCC tissues and adjacent nontumorous tissues. (B) The association between the expression of FoxQ1 and either VersicanV1 or macrophages in Cohort I HCC patients. (C1,C2) A Kaplan-Meier analysis of VersicanV1 expression or macrophage infiltration in Cohort I (n = 690) HCC patients after curative resection. (C3) The Kaplan-Meier analysis of concurrent FoxQ1 and VersicanV1 expression with recurrence and overall survival in Cohort I. (C4) The correlation of FoxQ1/macrophage coexpression with recurrence and overall survival in Cohort I. (D) The association between the expression of FoxQ1 and VersicanV1 or macrophage infiltration in Cohort II HCC patients. (E1,E2) A Kaplan-Meier analysis of VersicanV1 expression or macrophage infiltration in Cohort II (n = 312) HCC patients after curative resection. (E3) The correlation of FoxQ1/VersicanV1 coexpression with recurrence and overall survival in Cohort II. (E4) The correlation of FoxQ1/macrophage coexpression with recurrence and overall survival in Cohort II.

FoxQ1 Is Critical for β-Catenin-Mediated HCC Metastasis

The regulatory mechanism of FoxQ1 overexpression in HCC remains unknown. In this study we found that Wnt3a, constitutively active β-catenin, and TCF4 up-regulated FoxQ1 expression and transactivated its promoter activity (Supporting Fig. S3A-C). Serial deletion and site-directed mutagenesis showed that the induction of FoxQ1 expression by Wnt3a was dependent on 2 β-catenin/TCF4 binding sites within the FoxQ1 promoter (Supporting Fig. S3D-E). These studies suggest that Wnt/β-catenin signaling up-regulates FoxQ1 expression through both β-catenin/TCF4 binding sites within the FoxQ1 promoter. However, a recent study reported that β-catenin binds to only one TCF4/β-catenin binding site in the FoxQ1 promoter and increases transcription in colon cancer cells,[29] which is inconsistent with our study in HCC cells.

To determine whether FoxQ1 is involved in β-catenin-mediated metastasis, we down-regulated FoxQ1 expression in SMMC7721-β-catenin cells and up-regulated FoxQ1 expression in HCCLM3-shβ-catenin cells (Fig. 8A). Transwell assay showed that overexpression of β-catenin significantly increased the migration and invasion abilities of SMMC7721 cells, whereas down-regulation of FoxQ1 dramatically decreased β-catenin-induced cell migration and invasion. In contrast, knockdown of β-catenin decreased the migration and invasion abilities of HCCLM3 cells, where up-regulation of FoxQ1 rescued the decreased cell migration and invasion induced by β-catenin knocking down (Fig. 8B; Supporting Fig. S7). In addition, overexpression of β-catenin significantly increased the expression of FoxQ1, ZEB2, VersicanV1, and CCL2 in SMMC7721 cells, whereas down-regulation of FoxQ1 dramatically decreased the increased expression of ZEB2, VersicanV1, and CCL2 induced by β-catenin. In contrast, knockdown of β-catenin decreased the expression of FoxQ1, ZEB2, VersicanV1, and CCL2 in HCCLM3 cells, whereas up-regulation of FoxQ1 rescued the decreased expression of ZEB2, VersicanV1, and CCL2 induced by β-catenin knockdown (Supporting Fig. S6). An in vivo metastatic assay showed that the down-regulation of FoxQ1 decreased the incidence of lung metastasis and the number of metastatic lung nodules while increasing the overall survival time of the SMMC7721-β-catenin group. In contrast, the up-regulation of FoxQ1 rescued the decreased incidence of lung metastasis and the number of metastatic lung nodules while decreasing the overall survival time of the HCCLM3-sh β-catenin group (Fig. 8C1-C5).

Figure 8.

FoxQ1 is critical for β-catenin-mediated HCC metastasis. (A,B) Following the infection of the SMMC7721-β-catenin cells and HCCLM3-shβ-catenin cells with the lentivirus LV-shFoxQ1 or LV-FoxQ1, respectively, (A) the protein levels of β-catenin and FoxQ1 were detected by western blot, and (B) the cell migration and invasion capacities were assessed using transwell assays. (C) In vivo metastatic assay. (C1) Cell lines were transplanted into the livers of nude mice. Representative bioluminescent imaging (BLI) of the different groups is shown at 10 weeks following orthotopic implantation. (C2) The incidence of lung metastases in the different groups of nude mice. (C3) The overall survival of the nude mice in the different groups. (C4) The number of lung metastatic foci in each group. (C5) Representative H&E staining of the lungs from the different groups is shown. (D-F) FoxQ1 is positively correlated with elevated β-catenin expression in human HCC tissues, and their coexpression indicates poor prognosis. (D) Representative immunohistochemical images of β-catenin and FoxQ1 expression in HCC tissues and adjacent nontumorous tissues. (E1) The association between the expression of β-catenin and FoxQ1 in Cohort I HCC patients. (E2) Kaplan-Meier analysis of the correlation between β-catenin expression and the recurrence or overall survival of 690 HCC patients (Cohort I). (E3) Kaplan-Meier analysis of concurrent β-catenin and FoxQ1 expression with recurrence and overall survival in Cohort I. (F1) The association between the expression of β-catenin and FoxQ1 in Cohort II HCC patients. (F2) Kaplan-Meier analysis of the correlation between β-catenin expression and the recurrence or overall survival of 312 HCC patients (Cohort II). (F3) Kaplan-Meier analysis of concurrent β-catenin and FoxQ1 expression with recurrence and overall survival in Cohort II.

Immunohistochemical results in both independent cohorts showed that elevated β-catenin expression (both nuclear and cytoplasmic staining) was associated with poor prognosis (Fig. 8D,E2,F2) and aggressive tumor behavior (Supporting Table S9). Elevated β-catenin expression was positively correlated with FoxQ1 expression in human HCC tissues (Fig. 8E1,F1). Furthermore, patients with positive coexpression of β-catenin and FoxQ1 had the highest recurrence rates and lowest overall survival times (Fig. 8E3,F3). Taken together, these studies suggest that FoxQ1 is critical for β-catenin-mediated HCC metastasis.

To further explore the role of β-catenin-mediated FoxQ1 signaling pathway in HCC metastasis, we detected the protein levels of β-catenin, FoxQ1, ZEB2, E-cadherin, VersicanV1, and CCL2 in eight primary HCC tissues from patients who did not develop metastasis and eight primary HCC tissues from patients who developed metastasis. The expression of β-catenin, FoxQ1, ZEB2, VersicanV1, and CCL2 was much higher in primary HCC tissues from patients who developed metastasis than those who did not develop metastasis. However, the expression of E-cadherin was significantly lower in primary HCC tissues from patients who developed metastasis than those who did not develop metastasis (Supporting Fig. S8).

Discussion

In this study we found that FoxQ1 expression was significantly up-regulated in HCC tissues in two independent cohorts of human HCC patients. FoxQ1 overexpression was correlated with microvascular invasion, poor differentiation, and higher TNM stage. HCC patients with positive FoxQ1 expression had poorer prognoses than patients with negative FoxQ1 expression. A multivariate analysis revealed that FoxQ1 expression was an independent and significant risk factor for recurrence and survival. In addition, FoxQ1 expression was much higher in HCC tissues from patients who developed metastases than that in HCC tissues from patients who did not develop metastases. These clinical data strongly indicate that FoxQ1 contributes to the progression and metastasis of HCC.

To metastasize, cancer cells must attenuate cell-cell adhesion to disseminate into distant organs. Several transcription factors have been identified as critical regulator of this process, including Snai1, Slug, ZEB1, ZEB2/SIP1, Twist, and E49.[30] These proteins bind to E-box elements in the promoter region of E-cadherin, leading to the transcriptional inactivation of E-cadherin.[31] We previously reported that FoxC1 induces EMT by transactivating Snai1 expression.[32] In this study we found that FoxQ1 transactivated ZEB2 expression by directly binding to the ZEB2 promoter, thereby inhibiting E-cadherin transcription. The knockdown of ZEB2 expression significantly attenuated FoxQ1-enhanced invasion and lung metastasis, whereas the up-regulation of ZEB2 rescued the decreased invasion and lung metastasis induced by FoxQ1 knockdown. Furthermore, FoxQ1 expression was positively correlated with ZEB2 expression but inversely correlated with E-cadherin expression in human HCC tissues. The FoxQ1(+)/ZEB2(+) and FoxQ1(+)/E-cadherin(−) coexpression patterns were associated with poorer prognosis. Thus, FoxQ1-mediated ZEB2/E-cadherin signaling pathway plays an important role in promoting HCC metastasis and poor prognosis.

VersicanV1 promotes metastasis through the recruitment of macrophage infiltration, while the depletion of macrophages decreases VersicanV1-induced metastasis in bladder cancer and breast cancer.[16, 17] A recent study reported that the mRNA levels of VersicanV1 were significantly higher in HCC tissues than in normal liver tissues.[33] However, whether VersicanV1-mediated macrophage infiltration is involved in HCC metastasis has remained unknown. In this study we found that patients with positive VersicanV1 expression had shorter overall survival times and higher recurrence rates than patients with negative VersicanV1 expression; in addition, VersicanV1 overexpression was positively correlated with poor tumor differentiation and higher TNM stage. The up-regulation of VersicanV1 promoted HCC invasion and metastasis, whereas the down-regulation of VersicanV1 inhibited HCC invasion and metastasis. Furthermore, VersicanV1 expression in HCC cells promoted macrophage attraction in a CCL2-dependent manner. The overexpression of VersicanV1 in HCC cells significantly increased macrophage infiltration and inflammatory mediator expression. Thus, VersicanV1 expression in HCC cells promotes macrophage infiltration and HCC metastasis.

In this study we report a novel function for FoxQ1 in modifying the tumor microenvironment to promote HCC metastasis. Using serial deletion, site-directed mutagenesis, and ChIP assays, we found that VersicanV1 is a direct transcriptional target of FoxQ1. FoxQ1 expression in HCC cells promoted the migration and transendothelial migration abilities of macrophages through the VersicanV1/CCL2 axis. Depletion of macrophages dramatically decreased FoxQ1-induced HCC metastasis. These studies indicate that macrophage infiltration is critical for FoxQ1-mediated HCC metastasis.

In addition, VersicanV1-induced macrophage infiltration is involved in FoxQ1-mediated HCC metastasis. The inhibition of VersicanV1 expression decreased FoxQ1-induced macrophage infiltration, inflammatory mediator expression, and HCC metastasis. However, the up-regulation of VersicanV1 rescued the decreased macrophage infiltration, inflammatory mediator expression, and HCC metastasis induced by FoxQ1 knockdown. In two independent cohorts of human HCC tissues, FoxQ1 expression was positively correlated with VersicanV1 expression and intratumoral macrophage infiltration. The coexpression of either FoxQ1(+)/VersicanV1(+) or FoxQ1(+)/TAM(+) was associated with poorer prognosis. Thus, FoxQ1 promotes HCC metastasis by up-regulating VersicanV1 expression and recruiting macrophage infiltration.

In conclusion, we report a new molecular mechanism of FoxQ1 in HCC metastasis. FoxQ1 induces EMT and promotes HCC metastasis by transactivating ZEB2 expression. FoxQ1 expression in HCC cells promotes the recruitment of macrophage infiltration through the VersicanV1/CCL2 axis, which in turn promotes HCC metastasis. Thus, identifying the components of this pathway may provide potential therapeutic targets for the treatment of this deadly disease.

Ancillary