Hepatitis B virus X protein promotes the stem-like properties of OV6+ cancer cells in hepatocellular carcinoma

Hepatitis B virus X protein (HBx) and cancer stem-like cells (CSCs) have both been implicated in the occurrence and development of HBV-related hepatocellular carcinoma (HCC). However, whether HBx contributes to the stem-like properties of OV6+ CSCs in HCC remains elusive. In this study, we showed that the concomitant expression of HBx and OV6 was closely associated with the clinical outcomes and prognosis of patients with HBV-related HCC. HBx was required for the stem-like properties of OV6+ liver CSCs, including self-renewal, stem cell-associated gene expression, tumorigenicity and chemoresistance. Mechanistically, HBx enhanced expression of MDM2 by directly binding with MDM2 and inhibiting its ubiquitin-directed self-degradation. MDM2 translocation into the nucleus was also upregulated by HBx and resulted in enhanced transcriptional activity and expression of CXCL12 and CXCR4 independent of p53. This change in expression activated the Wnt/β-catenin pathway and promoted the stem-like properties of OV6+ liver CSCs. Furthermore, we observed that the expression of any two indicators from the HBx/MDM2/CXCR4/OV6 axis in HCC biopsies could predict the prognosis of patients with HBV-related HCC. Taken together, our findings indicate the functional role of HBx in regulating the stem-like properties of OV6+ CSCs in HCC through the MDM2/CXCL12/CXCR4/β-catenin signaling axis, and identify HBx, MDM2, CXCR4 and OV6 as a novel prognostic pathway and potential therapeutic targets for patients with HBV-related HCC patients.

Liver cancer is the second leading cause of cancer death in men worldwide. 1 Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer. 2 Chronic infection with hepatitis B virus (HBV) is a major risk factor for HCC. 3 Furthermore, hepatitis B virus X protein (HBx), encoded by HBV, is known to be necessary in the occurrence and development of HCC. 4 Our group has demonstrated that HBx promotes the expansion and tumorigenicity of hepatic progenitor cells (HPCs), which contribute to HBx-mediated tumor formation in a DDC-induced mouse model. 5 Accumulating evidence suggests that HBx modulates the transcriptional activities of many genes involved in cell survival and apoptosis by interacting with certain components of signal pathways such as Nrf2 and β-catenin. 6,7 In addition, HBx regulates the degradation of various proteins, including c-Myc, by interacting with components of the ubiquitin-proteasome system. 8 However, the molecular mechanisms involved in the association of HBx with HBV-related hepatocarcinogenesis remain elusive.
Although surgical resection and liver transplantation are curative treatments for HCC patients, most HCCs are prone to invasion and metastasis, and the long-term prognosis remains unsatisfactory. For patients in the advanced stages of HCC, transarterial chemoembolization (TACE) is the second-line treatment, but its overall effect is less than satisfactory due to high chemoresistance of the tumor. 9 Because of this, many reports have posited that cancer stem-like cells (CSCs) can be enriched by chemotherapeutic drugs, which is closely related to drug resistance and eventual relapse of HCC. 10,11 For instance, angiopoietin-like protein 1 (ANGPTL1) inhibits sorafenib resistance and cancer stemness in HCC cells by repressing EMT. 11 Therefore, it is essential to thoroughly understand the molecular mechanism underlying CSCs in the progression of HCC to develop more effective therapeutic strategies.
The CSC theory proposes that CSCs are defined as a subpopulation of cells in tumors that share similarities with normal stem cells, such as self-renewing capacity. The existence of CSCs was first reported in acute myeloid leukemia 12 and more recently in various solid tumors, including those of the brain, liver and breast. [13][14][15] Recently, CSCs in HCC have been identified by cell surface antigens such as CD133, EpCAM, CD24 and CD90. [16][17][18][19] Previous studies conducted by our group and others also identified OV6 as a marker for CSCs in HCC and demonstrated that OV6 + 1 cancer cells promoted HCC progression through the C-X-C chemokine receptor type 4 (CXCR4) and its specific ligand CXCL12 (SDF-1). 20 The CXCR4/CXCL12 pathway has been implicated in the invasive and stemness properties of cancer. 21 Upon CXCL12 binding, CXCR4 activates different pathways such as the PI3K/AKT and Wnt/β-catenin pathways, both of which contribute to the maintenance of the stem-like properties of CSCs. 22 However, there is still a limited understanding of the molecular mechanisms underlying CSCs in the invasion and metastasis of HCC.
In the present study, we demonstrated that concomitant HBx and OV6 expression was closely correlated to the HBx promotes expansion of liver CSCs C Wang et al clinicopathological characteristics and prognosis of HBVrelated HCC patients. Gene expression microarrays were utilized to demonstrate that MDM2 was the critical gene involved in HCC carcinogenesis. MDM2 is an oncogene originally identified as being amplified on double-minute chromosomes in transformed mouse fibroblasts. 23 MDM2 has been shown to have an important role in cancers, stem cells and cancer stem cells. [24][25][26] In addition, MDM2 is a member of the RING finger family of E3 ubiquitin ligases and negatively regulates the tumor suppressor protein p53. 27 However, recent studies have shown that MDM2 also exerts p53-independent activities. 28 In our model, MDM2 was shown to be necessary for HBx to regulate the stem-like characteristics of OV6 + liver CSCs in a p53-independent manner. Our findings also showed that the HBx/MDM2/CXCR4/OV6 axis could serve as a prognostic indicator for HBV-related HCC patients.

Results
Concomitant elevated expression of HBx and OV6 predicts a poor prognosis for patients with HBV-related HCC. Our previous results have demonstrated that HBx may promote the expansion and malignant transformation of HPCs, which contributes to HBx-driven tumorigenicity. 5 In this study, we further investigated the potential role of HBx in regulating putative liver CSCs. Our study identified OV6 as a novel biomarker of the liver CSC subpopulation in HCCs. 20 First, we determined the expression of HBx and OV6 in 267 patients with primary HCC using tissue microarrays (TMAs) (Figures 1a and b). A positive correlation between HBx and OV6 expression was observed in HBV-related HCC samples (Phi coefficient = 0.282; Po0.0001; Figure 1c). Based on the expression of HBx and OV6, all 267 patients were divided into four groups: I (n = 72), HBx high and OV6 high ; II (n = 63), HBx high and OV6 low ; III (n = 34), HBx low and OV6 high ; and IV (n = 98), HBx low and OV6 low (Table 1). In patients with HBx high HCC, the percentage with high OV6 staining was almost twice that of the HBx low group (53.33% versus 25.76%; Figure 1d). Moreover, compared with the other groups, the HBx high /OV6 high patients exhibited much more aggressive clinicopathological characteristics such as hepatitis B envelope antigen positivity (P = 0.021), tumor size (Po0.001), tumor number (P = 0.004) and microvascular invasion (Po0.001; Table 1). More importantly, the differences in both overall survival (OS) and disease-free survival (DFS) were significant among these four groups (Figures 1e and f). In addition, the multivariate analysis indicated that together with elevated OV6 levels and increased tumor numbers, upregulated HBx expression levels were an independent risk factor for both OS and DFS in HCC patients (Figure 1g). In conclusion, concomitant overexpression of HBx and OV6 can serve as an effective prognostic predictor for patients with HBV-related HCC.
HBx enhances the stem cell-like properties and tumorigenic potential of OV6 + liver CSCs. Next we explored whether overexpression of HBx could impact the stem-like characteristics of OV6 + liver CSCs. Using a lentivirus-based approach, we infected human HCC cells with lentivirus containing constructs expressing either HBx (LV-HBx) or empty vectors (LV-Con) and established HCC cells lines (SMMC-7721 and Huh7) that stably express HBx (Supplementary Figure S1). The flow cytometry results revealed that overexpression of HBx contributed to the expansion of the OV6 + population within the total HCC cell pool ( Figure 2a). Moreover, OV6 + cells magnetically sorted from LV-HBx HCC cell lines, exhibited increased stemnessrelated genes ( Figure 2b). Next we conducted in vitro spheroid formation assays to test whether HBx could regulate the self-renewal of liver CSCs. As expected, the OV6 + cells magnetically sorted from LV-HBx HCC cell lines were prone to form more tumor spheres when compared with OV6 + cells sorted from LV-Con HCC cell lines (Figure 2c). In addition, drug resistance is another important property of CSCs and may contribute to chemotherapeutic failure and tumor relapse. 29,30 To test whether HBx could manipulate the chemoresistance of liver CSCs, LV-HBx HCC cell lines were treated with chemotherapeutic agents commonly used in TACE treatment, including pirarubicin, oxaliplatin and hydroxycamptothecin (HCPT). The flow cytometry results revealed that HBx obviously enriched the chemo-resistant OV6 + subpopulation ( Figure 2d and Supplementary Figure 2A) and enhanced the cell survival ability of these cells after exposure to anticancer drugs ( Figure 2e and Supplementary Figure 2B), suggesting the critical role of HBx in regulating chemoresistance of OV6 + liver CSCs. Next, we performed xenograft assays using a gradient series of OV6 + subpopulations sorted from either HCC-LV-HBx or HCC-LV-Con cells. As shown in Figures 2f and g, subcutaneous injection of HBx-expressing OV6 + HCC cells resulted in an increased incidence of tumorigenicity, indicating that HBx overexpression could enhance the tumorigenicity of OV6 + subpopulations in vivo. Taken together, our findings (g) Multivariate analysis of the hazard ratios (HRs) for overall survival and tumor recurrence. Univariate Cox proportional hazard (PH) analyses were conducted for each variable and covariate. A multivariate Cox PH regression model was established, and the HRs were presented as the means (95% confidence interval). Only variables that met the PH assumption in the univariate analysis would be included in the multivariate Cox PH analysis suggest that HBx is required for the expansion and maintenance of putative OV6 + liver CSCs.
MDM2 is identified to be upregulated by HBx in OV6 + liver CSCs. To elucidate the underlying mechanisms by which HBx regulates the stem-like properties of OV6 + liver CSCs, we used gene expression microarrays to screen differentially expressed genes in OV6 + /LV-HBx HCC cells compared with the levels in LV-Con-infected OV6 + cells. By using a 2-fold cut-off with a P-value o0.05, a total of 544 genes were  hierarchical clustering heat map was used to show the distinguishable mRNA expression patterns among these samples ( Figure 3b). We next performed Gene Ontology (GO) and pathway analyses to determine the functional roles of these differentially expressed genes in GO terms and biological pathways, respectively. As shown in Supple-mentary Figure S3A, most of the enriched GO terms for biological processes targeted by dysregulated genes were closely related to the inflammatory response, the DNA HBx promotes expansion of liver CSCs C Wang et al damage response, and cell proliferation. A detailed pathway analysis revealed the enrichment of signaling pathways mainly corresponding to cell adhesion, the p53 signaling pathway, and cytokine receptor interactions (Supplementary Figure S3B). We next explored the possible biological interactions of these differentially regulated genes using Ingenuity Pathway Analysis (IPA). The highest-rated signal transduction network among the dysregulated genes in OV6 + /HBx-expressing cells was established. Intriguingly, the IPA analysis also identified MDM2 as the most highly scored upstream molecule in this network ( Figure 3c). Mounting evidence has shown that MDM2 is frequently overexpressed in the majority of malignances and promotes tumor development by negatively regulating the p53 pathway. [31][32][33] To substantiate our screening results, qRT-PCR and western blot assays were performed to verify the expression of MDM2 in sorted OV6 + cells expressing either HBx or the control vector. As expected, MDM2 were significantly upregulated in OV6 + /LV-HBx HCC cells when compared with those in the control cells (Figures 3d and e). Next, we verified this correlation in clinical tissues and performed IHC to determine the expression levels of MDM2 in HCC TMAs ( Figure 3f). As predicted, MDM2 expression was associated with HBx content in clinical HBV-related HCC biopsies (Phi coefficient = 0.595; Po0.0001; Supplementary Figure S3C). Similarly, a significant correlation between MDM2 and OV6 expression was also observed (Phi coefficient = 0.208; P = 0.001; Supplementary Figure S3D). The clinical data further revealed that HBV-related HCC patients with elevated HBx and MDM2 levels exhibited a worse OS and a shorter disease-free survival period than other patients ( Figure 3g and Supplementary Table S1). Meanwhile, concomitant overexpression of MDM2 and OV6 predicted rapid tumor relapse and a poor prognosis for HBV-related HCC patients (Figure 3h and Supplementary Table S2). Taken together, these data strongly suggest that HBx may promote the expansion of OV6 + liver CSCs in a MDM2-dependent manner.
MDM2 is required for HBx to promote the stem-like properties of OV6 + liver CSCs. To further verify the essential role of MDM2 in the HBx-induced expansion of OV6 + liver CSCs, we knocked down MDM2 expression in LV-HBx-infected HCC cells using small interfering RNA (siRNA). As shown in Figures 4a and b, both sequence 1 and sequence 3 elicited a stable knockdown effect. MDM2 has been recognized as a negative regulator of the tumor suppressor p53, which is involved in tumorigenesis and progression of cancers. 31,32 Therefore, we investigated whether the involvement of MDM2 in maintaining the HBxmediated stem-like properties of OV6 + CSCs was dependent on p53. Three HCC cell lines with different p53 expression states (HepG2, wild-type p53; Huh7, p53 mutant; and Hep3B, p53 deletion) were chosen for functional experiments. qRT-PCR analysis revealed that the mRNA levels of stemness-related genes were upregulated in sorted OV6 + /LV-HBx-infected HCC cells, whereas MDM2 silencing resulted in significantly decreased expression of these genes regardless of the p53 status ( Figure 4c). In addition, exogenous HBx expression enhanced the capacity of tumor sphere formation in OV6 + HCC cells, whereas downregulation of MDM2 abolished this effect ( Figure 4d). Next, we assessed whether depleting MDM2 could impair the HBx-mediated tumorigenic potential of OV6 + CSCs in vivo by using a limiting dilution assay. Compared with OV6 + cells sorted from HCC cells transduced with LV-Con, an equivalent number of OV6 + /LV-HBx HCC cells showed increased capability to form xenografts in NOD/SCID mice. Conversely, MDM2 knockdown decreased the tumorigenicity of OV6 + /LV-HBx cells in our xenograft model ( Figure 4e). Therefore, these data indicate that MDM2 promotes HBx-induced expansion of OV6 + liver CSCs in a p53-independent manner.
HBx upregulates the expression of MDM2 by inhibiting its self-ubiquitination and degradation. Because MDM2 is required for HBx to promote the stem-like characteristics of OV6 + liver CSCs, we examined the mechanisms underlying HBx-mediated MDM2 activity. First, we determined whether HBx could directly interact with MDM2. An immunofluorescence staining assay showed the co-localization of HBx and MDM2 in the cytoplasm and nuclei of OV6 + HCC cells (Figure 5a), suggesting the direct interaction between HBx and MDM2. In addition, the co-immunoprecipitation assay revealed that HBx and MDM2 strongly interacted with each other in OV6 + CSCs sorted from HBx-expressing HCC cells (Figure 5b). Since MDM2 is an unstable protein that can be ubiquitinated in an autocatalytic manner, we next examined whether HBx inhibited the self-ubiquitination of MDM2. Western blotting was conducted to demonstrate that HBx-  (Figure 5c). Therefore, these findings suggest that HBx regulated MDM2 degradation via a proteasomal-dependent pathway rather than by protein HBx promotes expansion of liver CSCs C Wang et al synthesis. As shown in Figure 5d, the ubiquitination of MDM2 was reduced when HBx was overexpressed in HCC cells. Taken together, HBx enhances expression of MDM2 by directly binding with MDM2 and inhibiting its selfubiquitination and degradation activities.
The CXCL12/CXCR4 autocrine loop is required for HBx/ MDM2 to mediate the stem-like properties of OV6 + liver CSCs. Our previous study showed that activation of CXCR4 maintained the self-renewal capacity of OV6 + HCC cells. 20 The CXCL12/CXCR4 pathway has been implicated in various types of cancers. 34 Both the autocrine and paracrine effects of this pathway have been shown to maintain cancer growth and progression. 35 Therefore, we investigated whether HBx/ MDM2 regulated the stemness of OV6 + CSCs through the CXCL12/CXCR4 pathway. First, CXCR4 expression was upregulated by HBx in OV6 + CSCs in HCC, whereas MDM2 knockdown reversed HBx-mediated CXCR4 upregulation (Figures 6a and b). Subsequently, the siRNA approach was used to inhibit the expression of CXCR4 in HCC cells (Supplementary Figure S4A). The number of spheres was greater in HBx-expressing OV6 + CSCs compared with control OV6 + CSCs, whereas CXCR4 knockdown eliminated HBxmediated self-renewal (Figure 6c). Similarly, inhibiting CXCR4 with its specific antagonist AMD3100 also attenuated the HBx-mediated self-renewal of OV6 + liver CSCs (Figure 6d). Furthermore, the qRT-PCR assay showed that CXCL12 expression was upregulated by HBx in OV6 + liver CSCs, but MDM2 depletion abolished this effect (Figure 6b). To examine the autocrine source of the CXCL12 ligand in cultured OV6 + CSCs, an ELISA assay was performed, which showed that CXCL12 protein was not found in control media (i.e, with no cells) compared with conditioned media from cultured OV6 + liver CSCs (Figure 6e). These data indicated that the in vitro source of CXCL12 protein is the OV6 + CSCs. Moreover, the protein levels of CXCL12 were increased in conditioned media from OV6 + CSCs overexpressing HBx and decreased in media from cultured OV6 + CSCs with HBx expression and MDM2 knockdown (Figure 6e). Furthermore, CXCL12 knockdown resulted in fewer spheres in HBxexpressing OV6 + CSCs, while the inhibited self-renewal was rescued by the addition of CXCL12 protein (Supplementary Figure S4B and Figure 6f). These data indicate that CXCL12 is secreted by OV6 + CSCs and is required in mediating the HBx/MDM2-induced stem-like properties of OV6 + liver CSCs. To further confirm that the effect of CXCL12 on the self-renewal of HBx-expressing OV6 + CSCs is dependent on CXCR4, we also added CXCL12 protein to the culture media of CXCR4-depleted HBx-expressing OV6 + CSC cells. As expected, we found no significant difference in the number of spheres after ligand stimulation compared with control cells (Figure 6f). However, inhibition of CXCR4 by AMD3100 significantly disrupted the effect of CXCL12 on the self-renewal capacity of HBx-expressing OV6 + CSCs (Figure 6g). In summary, these data demonstrate that the CXCL12/CXCR4 autocrine loop is critical in mediating the HBx/MDM2 induction of the stem-like properties of OV6 + liver CSCs.
HBx/MDM2-induced upregulation of CXCL12/CXCR4 activates the Wnt/β-catenin pathway in OV6 + liver CSCs. We next determined how HBx/MDM2 mediated CXCL12/CXCR4 expression. First, immunofluorescence staining showed that overexpression of HBx increased the amount of MDM2 localized in the nuclei of OV6 + liver CSCs (Figure 7a). Second, overexpression of HBx was demonstrated to enhance the transcriptional activation of CXCL12 and CXCR4 via luciferase reporter assays, whereas knockdown of MDM2 prevented HBx-induced transcriptional upregulation ( Figure 7b). Our previous study demonstrated that the CXCL12/CXCR4 axis promoted self-renewal and the migration of OV6 + cells in HCC. In addition, β-catenin has a vital role in the maintenance of OV6 + liver CSCs. Therefore, we asked whether CXCL12/CXCR4 enhanced the stem-like properties of OV6 + liver CSCs by activating the Wnt/β-catenin pathway. First, western blotting showed that CXCL12 enhanced the expression of β-catenin and cyclin D1, whereas CXCR4 depletion blocked the ability of CXCL12 to activate the Wnt/β-catenin pathway in HBx-expressing OV6 + liver CSCs (Figure 7c). Furthermore, the spheroid formation assay was performed to demonstrate that the number of spheres from HBx-expressing OV6 + liver CSCs treated with CXCL12 was greater than that of untreated HBx-expressing OV6 + CSCs. However, CXCR4 knockdown attenuated the effect of CXCL12 on enhancing tumor spheroid formation in HBx-expressing OV6 + CSCs (Figure 7d). Taken together, HBx enhances the transcriptional activation of CXCL12 and CXCR4, which activate the Wnt/β-catenin pathway in OV6 + liver CSCs.
HBx/MDM2/CXCR4/OV6 expression predicts malignant clinicopathological features and a poor prognosis for patients with HBV-related HCC. We also examined the clinical relationship among HBx/MDM2/CXCR4/OV6 expression in specimens from patients with HBV-related HCC. First, qRT-PCR and IHC analysis were conducted in HCC

Discussion
Hepatocellular carcinoma (HCC) comprises the majority of liver cancer cases, and the high HCC rates in China largely reflect the elevated prevalence of chronic hepatitis B virus (HBV) infection. 36 HBx has been demonstrated to have crucial roles in the relapse and metastasis of HCC. 37 Our previous study proved that HBx could promote the expansion and tumorigenicity of HPCs in DDC-treated mice. Importantly, we also observed there was a positive correlation between HBx and OV6 expression in samples from patients with HBV-related HCC. 5 These findings suggest that HBx may regulate the stem-like properties of OV6 + CSCs in HBV-related HCC. Consistent with this, other studies have reported that HBx contributes to hepatocarcinogenesis partially by promoting changes in gene expression that are characteristic of CSCs. 38,39 However, the molecular mechanisms underlying how HBx regulates the stem-like properties of CSCs remain elusive.
In this study, we demonstrated that HBx mediated the stemlike characteristics of OV6 + CSCs in HCC. A gene expression microarray demonstrated that MDM2 is upregulated in various types of tumors and especially increased in our model. Mounting evidence has demonstrated that MDM2 is a key negative regulator of the p53 tumor suppressor protein, which promotes tumor growth. 40 In addition to p53, a number of additional interaction partners for MDM2 have been described. 41 In our study, we demonstrated that HBx promoted the stem-like properties of OV6 + CSCs via MDM2 independent of p53. Other studies demonstrated that MDM2 had a crucial role in the stemness of CSCs from breast cancer and glioblastoma. 42,43 It is well established that MDM2 is an unstable member of the RING finger family of E3 ubiquitin ligases that is ubiquitinated in an autocatalytic manner. Our results demonstrated that HBx enhanced the expression of MDM2 by directly binding to MDM2 and inhibiting its selfdegradation.
Subsequently, we presented evidence that HBx promoted increased translocation of MDM2 into the nucleus, which enhanced the transcriptional activation and expression of  Figure 6 The CXCR4/CXCL12 autocrine loop is required in HBx/MDM2 mediation of the stem-like properties of OV6 + CSCs. (a) Western blotting was performed to assess the expression of MDM2, CXCR4 and Flag-HBx in OV6 + CSCs, OV6 + CSCs that overexpressed Flag-HBx and OV6 + CSCs that overexpressed Flag-HBx and MDM2 siRNA, which were magnetically isolated from HepG2, Huh7 and Hep3B cells, respectively. (b) Quantitative RT-PCR analysis was used to assess the CXCR4 and CXCL12 expression in OV6 + CSCs, OV6 + CSCs that overexpressed Flag-HBx and OV6 + CSCs that overexpressed Flag-HBx and MDM2 siRNA, which were isolated from HepG2 and Huh7 cells (*Po0.05). (c and d) Primary and secondary passages of HBx-expressing OV6 + CSCs from the HepG2 and Huh7 cell lines generated more spheroids. When MDM2 was knocked down in HBx-expressing OV6 + CSCs by siRNAs (siCXCR4, sequence #2 and #3) or AMD3100 treatment (20 μg/ml), number of spheres was reduced. Experiments were performed in triplicate and all data represent mean ± S.D., *Po0.05, **Po0.01 and ***Po0.001. (e) ELISA assay was performed to measure the CXCL12 protein level in PBS, media alone (DMEM) and conditioned media from OV6 + CSCs and HBx-expressing OV6 + CSCs transfected with either control siRNA or siRNA targeting MDM2 (*Po0.05, **Po0.01 and ***Po0.001). (f) The spheroid assay was used to assess number of spheres of HBx-expressing OV6 + CSCs transfected with control siRNA (siCtrl), siRNA targeting CXCL12 (sequence #3), or treated with recombinant CXCL12 protein plus siRNA targeting CXCL12, or recombinant CXCL12 protein plus siRNA targeting CXCR4. (g) The spheroid assay was conducted to assess the number of spheres of HBx-expressing OV6 + CSCs treated with recombinant CXCL12 protein alone, or recombinant CXCL12 protein plus AMD3100 at a concentration of 20 μg/ml. In f and g, experiments were performed in triplicate. Representative results from three independent experiments are shown, and all data are presented as the mean ± S.D., *Po0.05, **Po0.01 and ***Po0.001 CXCL12 and CXCR4 in OV6 + liver CSCs. Because our previous study indicated that CXCL12/CXCR4 was necessary to maintain the stemness of OV6 + CSCs in HCC, 20 we next assessed whether HBx/MDM2 regulated CXCL12/CXCR4 signaling. As expected, our data indicate that the CXCL12/ CXCR4 autocrine loop is critical in HBx/MDM2 mediating the stem-like properties of OV6 + CSCs. We also proved that HBx/ MDM2 regulation of CXCL12/CXCR4 signaling resulted in activation of the Wnt/β-catenin pathway in OV6 + liver CSCs.
In conclusion, HBx induces greater stem-like features of OV6 + CSCs in HCC via the MDM2/CXCL12/CXCR4/β-catenin signaling axis independent of p53. Furthermore, increased expression of any two indicators among HBX/MDM2/CXCR4/ OV6 could predict the clinicopathological characteristics and prognosis of patients with HBV-related HCC. The related molecular mechanism will be discussed in depth in our future work.

Materials and Methods
Clinical HCC tissue microassays. Ethic approval of this study was obtained from the Ethic Committee Board of Eastern Hepatobiliary Surgery Hospital (Shanghai, China), and informed consent was obtained from each patient. The HCC tissue microarrays containing 267 patients who underwent curative hepatectomy for HCC at our hospital were constructed and analyzed according to previous instructions. 44 The clinicopathological features of patients were summarized in Supplementary Table S4. All the patients were followed until September 2013, with a median observation time of 54 months. Overall survival (OS) represented the interval between the date of surgery and death. Disease-free survival (DFS) was defined as the interval between the dates of surgery and recurrence. Quantitative real-time PCR. Total RNA from HCC cell lines and human HCC specimens were extracted using TRIzol reagent (Gibco) based on the manufacturer's protocols. Quantitative RT-PCR assays were carried out using an ABI 7300 Fast Real-Time PCR System (Applied Biosystems, Carlsbad, CA, USA) Figure 7 HBx/MDM2-induced upregulation of CXCL12/CXCR4 activates the Wnt/β-catenin pathway in OV6 + liver CSCs. (a) Immunofluorescence assays were performed using the OV6 + HCC cells (HepG2 and Huh7) infected with either LV-HBx or LV-Con. The intracellular localization of MDM2 (red) was detected by confocal laser scanning microscopy as indicated. Nucleus was stained by DAPI dye (blue; scale bar = 10 μm). (b) OV6 + LV-Con and LV-HBx HCC cells transfected with either negative control siRNA or siMDM2 were transfected with CXCL12 and CXCR4 luciferase reporters, followed by luciferase assay. The data display the mean ± S.D., *Po0.05 and **Po0.01. (c) HBxoverexpressing HCC cells treated with control vehicle, recombinant CXCL12 protein alone or recombinant CXCL12 protein plus siRNA targeting CXCR4 for 48 h before collecting. The expression levels of indicated molecules in whole-cell lysates were detected by western blotting. β-actin was used as internal loading control. (d) The spheroid assay was performed to assess the number of spheres of HBx-expressing OV6 + CSCs treated with control vehicle, recombinant CXCL12 protein alone or recombinant CXCL12 protein plus siRNA targeting β-catenin. Experiments were performed in triplicate and all data represent the mean ± S.D., *Po0.05, **Po0.01 and ***Po0.001 and SYBR Green PCR kit (Applied TaKaRa, Shiga, Japan). The primer sequences listed in Supplementary Table S6 were designed and purchased from Intergrated Biotech Solutions, Co., Ltd. Each measurement was performed in triplicate and the results were normalized by the expression of the β-actin gene. Fold change relative to mean value was determined by 2 − ΔΔCt .
Western blotting and co-immunoprecipitation. Whole-cell extracts were prepared in lysis buffer and western blotting assays were performed as described previously. 45 Briefly, cultured cells or liver cancer tissues were lysed in Triton lysis buffer (20 mM Tris, pH 7.4, 137 mM NaCl, 10% glycerol, 1% Triton X-100, 2 mM EDTA, 1 mM PMSF, 10 mM sodium fluoride, 5 mg/ml of aprotinin, 20 mM leupeptin and 1 mM sodium orthovanadate) and centrifuged at 12 000 × g for 15 min. Protein concentrations were determined via the BCA assay kit according to the manufacturer's protocols. Specific primary antibodies were shown as following: anti-MDM2 and anti-β-actin (Santa Cruz Biotechnology, Dallas, TX, USA), anti-CXCR4 (Abcam, Cambridge, MA, USA), anti-Flag-tag, anti-HA-tag, anti-MYCtag, cyclin D1 and anti-β-catenin (Cell Signaling Technology, Danvers, MA, USA). The immunocomplexes were incubated with the fluorescein conjugated secondary antibody and then analyzed using an Odyssey fluorescence scanner and LI-COR imaging system (Li-COR Biosciences, Lincoln, NE, USA). For immunoprecipitation experiments, a total of 1 mg of HCC cell lysates were incubated with 2 μg anti-Flag, or 2 μg anti-HA antibodies or normal rabbit immunoglobulin G (Santa Cruz Biotechnology) for 8 h at 4°C, followed by addition of Protein A/G Plus-Agarose (Santa Cruz Biotechnology) for another 3 h. The samples were denatured and subjected to western blotting analyses.
In vitro ubiquitination analysis. To determine ubiquitin-conjugated endogenous MDM2, HA-MDM2 and MYC-Ub plus Flag-HBx plasmids were cotransfected into 293T cells using jetPEI (Polyplus, New York, NY, USA) for 48 h before collecting. Whole-cell extracts were prepared in lysis buffer, as mentioned above. The cell lysates were incubated with HA-conjugated beads (Invitrogen) overnight at 4°C. Immunoprecipitated proteins were analyzed by western blotting using an anti-MYC antibody.
Immunohistochemistry. HCC tissue microarray slides were stained with the following primary antibodies at 4°C overnight: anti-OV6 (R&D Systems), anti-MDM2 (Santa Cruz Biotechnology) and anti-CXCR4 (Abcam). Corresponding secondary antibodies and diaminobenzidine (DAB) reagent (Dako, Carpinteria, CA, USA) were used in the detection procedure. All TMA slides were observed and photographed with an Olympus microscope (IX-70 OLYMPUS, Shinjuku, Tokyo, Japan). The staining levels of OV6, MDM2 and CXCR4 in all clinical samples were examined by two independent observers as described previously. 20 The staining intensities were scored as 0 (negative), 1 (weakly positive), 2 (moderately positive) or 3 (strongly positive). Low expression level was defined as score ⩽ 1, whereas high expression in tumoral tissues was identified as score ⩾ 2.
Flow cytometry and magnetic cell sorting. The magnetic-activated cell sorting and flow cytometric assay were carried out by using a Moflo XDP Cell Sorter (Beckman, Boulevard Brea, CA, USA). HCC cell lines were magnetically labeled with PE-conjugated-OV6 antibody (mouse IgG1; Miltenyi Biotec, Auburn, CA, USA) and subsequently incubated with rat anti-mouse IgG1 micro-Beads, and then separated on MACS MS column (Miltenyi Biotec). All the procedures were performed according to the manufacturer's instructions. HCC cells infected with LV-Con or LV-HBx were incubated with OV6 antibody and the percentages of OV6 + cells were measured by flow cytometry. For chemotherapy resistance experiments, HCC cells that stably expressed vector or HBx were treated with pirarubicin, oxaliplatin or HCPT for 48 h and the proportions of OV6 + cells were quantified using flow cytometry as mentioned above. For cell sorting experiments, OV6 + cell populations were magnetically sorted from HCC cells that stably expressed HBx or empty vectors and then subjected to in vitro functional studies or in vivo tumorigenicity experiments.
Spheroid formation assay. For HCC cell lines, single-cell suspensions of 3000 magnetic sorted OV6 + cells were maintained in six-well ultra-low attachment culture microplates (Corning, Steuben County, NY, USA) for 14 days. For the second generation of sphere formation, formed spheroids were re-trypsinized to single-cell suspensions and equivalent number of isolated OV6 + cells were resuspended in low attachment cultural conditions for another 14 days. The spheroid number was counted under a microscopy 14 days after seeding and the representative pictures were taken.
Cell viability assay. The cell viability assay was performed with cholecystokinin-8 (CCK-8; Dojindo, Japan) as previously reported. 46 Briefly, 6000 viable cells were seeded in triplicates in 96-well plates and treated with pirarubicin, oxaliplatin or HCPT since the onset of culture period for 4 days. At indicated time points, each well was added with 10 μl CCK-8 and incubated for another 1 h. Then, the OD values were determined via a microplate reader (Synergy HT, BioTek, Winooski, VT, USA) at an absorbance of 450 nm. The cell viability rates were presented as a percentage of the control value which was recorded at the beginning (Time 0).
Luciferase reporter assay. A dual-luciferase reporter assay was performed by using wild-type TCF-luciferase construct (pGL3-OT) and the mutant TCFluciferase reporter construct (pGL3-OF). The promoter region of CXCR4 and CXCL12 were amplified by PCR. The pGL3-CXCR4 or CXCL12-luciferase plasmids were constructed by inserting these PCR products into the pGL3-Report Dual-Luciferase reporter vector (Ribobio Co., Guangzhou, China). Indicated HCC cells were cultured in 24-well plates. After incubating for 24 h, each well was transfected by lipofectamine 2000 (Invitrogen) with 500ng/well CXCR4 or CXCL12-luciferase plasmids with jetPEI reagents (Polyplus). Meanwhile, all wells were cotransfected with pRL-TK plasmids which contained the Renilla luciferase genes as internal control for normalizing transfection efficiency. Then the luciferase activity was determined using the Promega luciferase assay reagent and Synergy 2 Multi-Detection Microplate Reader after 48 h of incubation. The relative transcriptional activities of CXCL12 and CXCR4 were presented as the OT/OF ratio. 47 Microarray analysis of gene expression. The gene expression profiles were determined by Agilent human cDNA Microarray assay (Agilent Technologies, Santa Clara, CA, USA). Total RNA was extracted from OV6 + /LV-Con and OV6 + /LV-HBx Huh7 cells expressing using an RNeasy Mini kit (Qiagen, Hilden, Germany). Then, 10 μg of total RNA was reverse-transcribed to produce biotin-labeled cRNA and fragmented by adding Blocking Agent and Fragmentation Buffer. Then the hybridization solution was dispensed into the gasket slide and assembled to the mRNA expression microarray slide. The slides were incubated for 17 h at 65°C in an Agilent Hybridization Oven. The hybridized arrays were washed, fixed and scanned with using the Agilent DNA Microarray Scanner (part number G2505C). Agilent Feature Extraction software (version 11.0.1.1) was used to analyze acquired array images. Quantile normalization and subsequent data processing were performed with using the GeneSpring GX v11.5.1 software package (Agilent Technologies). After quantile normalization of the raw data, identified mRNAs were chosen for further data analysis. The detected signals were evaluated by gene hierarchical clustering of logarithmic values. Cluster 3.0 and Java TreeView 1.0.13 software were applied to generate a dendrogram for each cluster of genes according to their expression profiles.
GO and ingenuity pathway analysis. The GO analysis functionally assessed the differentially expressed genes with GO categories which were derived from Gene Ontology. The ontologic pathway enrichment analyses for the differently expressed genes were performed with particular attention to GO biological processes. P-values less than 0.05 were considered significant. Pathway analysis was applied to identify the significant pathways based on the KEGG database. The inter-regulation network analysis was carried out by BioGenius Co. (Shanghai, China). For ingenuity pathway analysis (IPA), the identified genes with altered expression profiles from array assays were imported into the IPA Tool (Ingenuity H Systems, Redwood City, CA, USA). Then the biological networks, global functions and functional pathways of a specific data set were recognized by the IPA analyses.
In vivo xenograft assay. Six-week-old male nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice were purchased from Chinese Science Academy (Shanghai, China). All animal experiments were carried out according to the Second Military Medical University Animal Care Facility and the National Institutes of Health guidelines. OV6 + HCC cells expressing vector or HBx were subcutaneously injected into right flank of mice at different cell numbers as indicated. The proportion of OV6 + CSCs was evaluated by an in vivo limiting dilution assay according to the manufacturer's instructions.
ELISA assay. Human CXCL12/SDF-1 ELISA Kit (BioVision, Milpitas, CA, USA) was applied in our ELISA assays. All procedures are in accordance with the manufacturer's instructions. Briefly, bring all the reagents and samples to room temperature before use. Add 100 μl of each standard and sample into appropriate wells. Cover well and incubate for 2.5 h at room temperature or overnight at 4°C with gentle shaking. Discard the solution and wash four times with 1 × Wash Solution. Add 100 μl of 1 × prepared biotinylated antibody to each well. Incubate for 1 h at room temperature with gentle shaking. Discard the solution. Repeat the wash steps. Add 100 μl of prepared Streptavidin solution to each well. Incubate for 45 min at room temperature with gentle shaking. Discard the solution. Repeat the wash steps. Add 100 μl of TMB One-Step Substrate Reagent to each well. Incubate for 30 min at room temperature in the dark with gentle shaking. Add 50 μl of Stop Solution to each well. Read at 450 nm immediately and finally calculate the related value.