MicroRNA-200a suppresses metastatic potential of side population cells in human hepatocellular carcinoma by decreasing ZEB2

Although microRNA-200a (miR-200a) is frequently downregulated in cancer, its role in side population (SP) has not been investigated. In this study, 101 pairs of primary hepatocellular carcinoma (HCC) tissues and matched normal control tissues were analyzed for miR-200a expression and its clinicopathological value was determined. We found that miR-200a was downregulated in HCC/SP and this was associated metastasis. MiR-200a suppressed metastasis of SP cells. Overexpression of miR-200a in SP cells decreased metastasis-related markers and expression of ZEB2. The associations between miR-200a, SP cells and ZEB2 were validated in HCC. These findings reveal that miR-200a suppresses metastasis of SP cells by downregulating ZEB2.


INTRODUCTION
Metastasis is a major cause of death in patients with hepatocellular carcinoma (HCC), although survival has improved due to advances in surgical techniques [1].
Cancer stem cells (CSCs) are characterized by their capacity for indefinite self-renew and by their relative quiescence [2,3]. In several types of cancer, side populations (SPs) have been shown to be enriched for cells with CSC-like activity and a CSC phenotype [4]. Recent studies have suggested that SP cells in human pancreatic cancer are characteristically chemoresistant [5]. Furthermore, a recent study reported that SP cells possess the ability to self-renew, and to divide asymmetrically into SP and non-SP cells [6]. Other reports have indicated that SP cells may serve as an ideal model for stem cell for research [7].
MicroRNAs (miRNAs) are frequently dysregulated in cancer where they may behave as tumor suppressor genes [8,9] or oncogenes [10]. Recently, it was reported that miR-200a plays a crucial role in the development of cancer through its regulation of epithelial to mesenchymal transition (EMT), cell migration, proliferation and metastasis [11,12,13]. In breast and ovarian cancer, the down-regulation of miR-200a plays an important role in cancer metastasis [14,15,16]. EMT is a process by which epithelial cells lose their cell-cell adhesion and gain invasive properties, which leads to the acquisition of mesenchymal stem cell characteristics [17,18]. Previously, it was shown that mesenchymal to epithelial transition (MET) is promoted by repression of the zinc finger E-boxbinding 1 (ZEB1) and ZEB2 expression [19,20]. In our previous study, global miRNA expression profiles of SP cells in F344 rat HCC cells and fetal liver cells were screened through a microarray platform. We found that miR-200a was significantly downregulated in the SP cells of HCC compared with the fetal liver cells [21].
In this study, we show that low expression of miR-200a promotes human HCC SP cells to metastasize through the transactivation of ZEB2 expression, which results in the induction of EMT. www.impactjournals.com/oncotarget

Downregulation of miR-200a is associated with metastasis of HCC and poor prognosis
MiR-200a expression was analyzed in primary tumor specimens from 101 patients with HCC. The gene expression level of miR-200a was 0.59±0.08 in the tumor specimens and 1.40±0.41 in the corresponding non-tumor tissues (Fig. 1A). The index of miR-200a expression in the HCC cell lines MHCC-97H, HepG2, Huh-7 and SMMC-7721 was significantly lower than in the normal human hepatocyte cell line HL-7702 (Fig. 1B). We compared the low (51 cases) and high (50 cases) miR-200a expression groups ( Table 1) and found that low expression of miR-200a strongly correlated with metastasis (P=0.011). The level of miR-200a expression was significantly lower in 74 cases of primary tumors with clinically confirmed metastasis compared with the 27 cases without metastasis (Fig. 1C).
The Kaplan-Meier analysis revealed that low miR-200a expression was associated with a shorter overall survival (P<0.01) (Fig. 1D). The multivariate Cox regression analysis revealed that low expression of miR-200a (P<0.001) as well as differentiation and metastasis were independent prognostic factors of patient survival (

MiR-200a is inversely correlated with ZEB2 expression but is positively correlated with E-cadherin expression
In our previous studies, we found that miR-200a inhibits the EMT process via the activation of transcriptional factors such as ZEB2, which is required during EMT; this has been thoroughly demonstrated in rat hepatic oval cells [22]. We further investigated   the possible relationship between miR-200a, ZEB2 and E-cadherin expression in human HCC tissues. Immunohistochemistry revealed that miR-200a expression was inversely correlated with ZEB2 expression but was positively correlated with E-cadherin expression ( Fig.  2A,B).

Identification of side population
The subpopulation is shown in a representative sample (red color-nuclear staining with Hoechst33342 dye; SP-side population; NSP-non-side population). (Fig.  3A). HCC cells were analyzed by dual wavelength FACS after incubation with Hoechst 33342. Representative results that were analyzed by flow cytometry are shown (0.1% in HepG2, 2.8% in MHCC-97H, 0.9% in SMMC-7721, 1.3% in Huh7) (Fig. 3B). To understand the influence of differentiation in this subpopulation, SP cells were detected as a subpopulation of SP and NSP cells after 4 days in culture (37% and 1.7% in MHCC-97H, 41% and 0.9% in Huh7, respectively) (Fig. 3C).

Characteristics of the subpopulation
ALBU, a marker of mature hepatocytes, was lower in SP cells and higher in non-SP cells. Conversely, KRT14, a marker of liver stem cells, was highly expressed in SP cells but was weakly expressed in NSP cells (Fig. 4A). The number of spheres formed by SP cells from the Huh7 and MHCC-97H cell lines was significantly higher than the corresponding number of NSP cells. In addition, the SP cells were able to induce larger spheres than the non-SP cells (Fig. 4B).
The invasion and migration abilities of MHCC-97H and Huh7 cells were higher in SP cells than in NSP cells (Fig. 4C). The morphologic appearance of SP and NSP cells was determined after staining for F-actin filaments and after the nuclei were stained with Rhodamine Phalloidin and DAPI. Under fluorescence microscopy after different times in culture (either 8 h or 72 h after the subpopulation cells were sorted), the SP cells of the Huh7 cell line appeared confluent with little cytoplasm and large nucleolus, while the NSP cells demonstrated expanded cytoplasm and a considerably small pyknotic nucleolus (Fig. 4D).

MiR-200a inhibits the metastasis of SP cells in vitro
The index of miR-200a expression in SP cells of MHCC-97H and Huh7 was significantly lower than that in NSP cells from both the MHCC-97H and Huh7 cell lines (Fig. 5A). Expression of miR-200a after transfection was confirmed by qRT-PCR (Fig. 5B). Expression of metastasis-related markers was confirmed by qRT-PCR after transfection (Fig. 5C). E-cadherin and ZO-1 were over-expressed in the MHCC-97H-SP-miR-200a group but were weakly-expressed in the Huh7-SP-KD-miR-200a group. In contrast, the expression values of N-cadherin, VASP, LAMC2 and ZEB2 were opposite to those observed for the other above-mentioned genes. The upregulation of miR-200a expression decreased the ability of the MHCC-97H-SP cells to invade and migrate. Conversely, the inhibition of miR-200a expression in Huh7-SP cells enhanced the ability of the SP cells to invade and migrate (Fig. 5D).

MiR-200a inhibits the metastasis of SP cells in vivo
SP cells were injected into the caudal vein of nude mice. Representative bioluminescent imaging (BLI) at 8 weeks is shown for the different groups (Fig. 6A). The in vivo metastatic assay showed that the upregulation of miR-200a decreased the incidence of lung metastasis and the number of metastatic lung nodules on the surface, but increased the overall survival time in the MHCC-97H-SP-miR-200a group. In contrast, the down-regulation of miR-200a increased the incidence of lung metastasis and the number of metastatic lung nodules on the surface but decreased the overall survival time in the Huh7-SP-KD-miR-200a group (Fig. 6B,C,D,F). Representative hematoxylin and eosin (H&E) staining showed that metastatic nodules were observed in the lungs of mice in the MHCC-97H-SP-Control group. There were no obvious metastatic nodules in the lungs of mice in the MHCC-97H-SP-miR-200a group. In contrast, metastatic nodules were observed in the lungs of mice in the Huh7-SP-KD-miR-200a group, whereas no metastatic nodules were found in mice in the Huh7-SP-KD-Control group (Fig. 6E). In addition, immunohistochemistry revealed that miR-200a expression was inversely correlated with ZEB2 expression but was positively correlated with E-cadherin expression (Fig. 6G).

MiR-200a induces the metastasis of SP cells through the transactivation of ZEB2 expression
The target site of miR-200a in the 3'UTR of ZEB2 was predicted (Fig. 7A). We also over-expressed or downregulated the miR-200a expression in MHCC-97H, and by western-blot we determined that miR-200a had the opposite function of ZEB2 (Fig. 7B). A luciferase reporter gene assay showed that, compared with the NC group, miR-200a mimics significantly inhibited the 3'UTR wild-type (WT) ZEB2 activity (P<0.01) and a miR-200a inhibitor significantly increased its activity (P<0.01). MiR-200a mimics and a miR-200a inhibitor had no effect on the activity of 3'UTR mutant (MT) ZEB2 (Fig. 7C) (Y-axis signifies the luciferase activity of ZEB2).
MiR-200a decreased ZEB2 and N-cadherin expression and increased E-cadherin and ZO1 expression in the MHCC-97H-SP-miR-200a group. The upregulation of ZEB2 expression inhibited the increase in E-cadherin expression, ZO1 expression as well as the loss of N-cadherin expression induced by miR-200a. In contrast, the knockdown of miR-200a increased ZEB2 and N-cadherin expression and decreased E-cadherin and ZO1 expression in the Huh7-SP-KD-miR-200a group. The inhibition of ZEB2 dramatically attenuated the loss of E-cadherin expression, ZO1 expression and the increase in N-cadherin expression (Figs. 5C,7D,E). Similar results were obtained in a western-blot analysis (Fig. 7F). The  down-regulation of ZEB2 significantly reduced the migration and invasion abilities of the cells, which was induced by miR-200a knockdown. The up-regulation of ZEB2 rescued the miR-200a-induced decrease in cell migration and invasion (Fig. 7G).

DISCUSSION
MiRNA expression profiling has shown that miR-200a is frequently downregulated [23,24,25,26]. In this study we found that miR-200a expression was significantly downregulated in SP cells of human HCC cell lines and HCC tissues. As in the case with HCC cell lines, miR-200a expression was much lower in HCC tissues from patients who developed metastasis than in patients who did not develop metastasis. In addition, these clinical data strongly indicate that low expression of miR-200a contributes to the metastasis of HCC.
Recently, SP cells were isolated from the HeLa cell line and shared morphological characteristics with stem-like cells, including a rounder shape, smaller size and higher adherence, unlike the NSP cells [27]. In our research, ALBU was lower in SP cells but higher in NSP cells. Conversely, the expression of KRT14 was high in SP cells but was low in NSP cells. Additionally, SP cells had higher capacity for sphere formation and a greater capacity for invasion and migration than NSP cells. These results suggest that SP cells possess certain stem cell characteristics. In the last study, researchers primarily focused on functional studies of SP cells instead of on molecular mechanistic studies. In this study, we found that the down-regulation of miR-200a may promote the metastasis of SP cells, whereas the up-regulation of miR-200a inhibited the metastasis of SP cells.
EMT is driven by ZEB, SNAIL, AKT2 and Forkhead Box (FOX) transcription factors that activate genes associated with a mesenchymal phenotype and that repress epithelial marker genes [28,29,30,31,32]. These proteins bind to the promoter regions, which leads to the transcriptional inactivation of E-cadherin, ZO1 or the activation of N-cadherin and Vimentin [33,34]. In this study, we found that miR-200a inhibited ZEB2 expression by binding to the ZEB2 promoter, which inhibits epithelial to mesenchymal transition. MiR-200a decreased N-cadherin expression and increased E-cadherin expression. The up-regulation of ZEB2 expression inhibited the increase in E-cadherin expression and the loss of N-cadherin expression. Thus, the miR-200amediated ZEB2/EMT signaling pathway is essential for SP cells in HCC cell lines to metastasize.
In conclusion, this study delineates the function of miR-200a in SP cells of HCC. Thus, further studies of SP cells from HCC and the components of this pathway may provide new insights into the treatment of this deadly disease.

Flow cytometry
The cells were labeled with Hoechst 33342 (Sigma, USA) at a concentration of 5 ug/ml for SMMC-7721, Huh7 and 8 ug/ml for MHCC-97H. The cells were incubated for 90 minutes at 37°C in the absence or presence of 60 ug/ml verapamil (Sigma, USA). After staining, the cells were suspended in cold DMEM that contained 2 ug/ml propidium iodide (PI, Sigma, USA), 2% FBS and 10 mM HEPES, then passed through a 40 um mesh filter (BD,USA) and maintained at 4°C until flow cytometry analysis.

In vivo metastasis assay
Male athymic nude mice (4-6 weeks old) were used. SP cells (1×104) were stably transfected with the luciferase gene, suspended in DMEM and injected into the caudal vein of nude mice. The location of tumors in the nude mice was detected by a bioluminescent imaging (BLI) system as previously described [35].

Cell transfection
The human miR-200a gene was constructed in GV254 (Genechem, China) and designated as either LV-miR-200a or LV-knockdown-miR-200a (LV-KD-miR-200a). An empty vector was used as the negative control and was designated LV-control or LV-KD-control. ZEB2 siRNA and negative controls (NC) were obtained from Genechem Co. (China) and were designated SP-ZEB2 or SP-anti-ZEB2.

Statistical analysis
Data were analyzed with SPSS 17.0 software. To evaluate significant differences between two matched paired groups or between two independent groups of samples, paired t tests and the Mann-Whitney U test were used, respectively. Survival curves were plotted according to the Kaplan-Meier method and were compared by the log-rank test. Data are expressed as the mean ± SD and P< 0.05 was considered statistically significant.

ACKNOWLEDGMENTS
We thank Fuqin Zhang (the Xijing hospital of the Fourth Military Medical University) for substantial technical assistance, Jing Qian (Becton, Dickinson and Company) and the Burn Departmen (the Xijing hospital