Downregulation of microRNA-15b by hepatitis B virus X enhances hepatocellular carcinoma proliferation via fucosyltransferase 2-induced Globo H expression

Authors

  • Chen-Shiou Wu,

    1. The Ph.D. Program for Cancer Biology and Drug Discovery, China Medical University, Taichung, Taiwan
    2. Graduate Institute of Cancer Biology and Center for Molecular Medicine, China Medical University, Taichung, Taiwan
    3. Genomics Research Center, Academia Sinica, Taipei, Taiwan
    Search for more papers by this author
  • Chia-Jui Yen,

    1. Graduate Institute of Clinical Medicine, National Cheng Kung University, Tainan, Taiwan
    2. Division of Hematology/Oncology, Department of Internal Medicine, National Cheng Kung University Hospital, Tainan, Taiwan
    Search for more papers by this author
  • Ruey-Hwang Chou,

    1. The Ph.D. Program for Cancer Biology and Drug Discovery, China Medical University, Taichung, Taiwan
    2. Graduate Institute of Cancer Biology and Center for Molecular Medicine, China Medical University, Taichung, Taiwan
    3. Department of Biotechnology, Asia University, Taichung, Taiwan
    Search for more papers by this author
  • Jia-Ni Chen,

    1. The Ph.D. Program for Cancer Biology and Drug Discovery, China Medical University, Taichung, Taiwan
    2. Graduate Institute of Cancer Biology and Center for Molecular Medicine, China Medical University, Taichung, Taiwan
    3. Genomics Research Center, Academia Sinica, Taipei, Taiwan
    Search for more papers by this author
  • Wei-Chien Huang,

    1. The Ph.D. Program for Cancer Biology and Drug Discovery, China Medical University, Taichung, Taiwan
    2. Graduate Institute of Cancer Biology and Center for Molecular Medicine, China Medical University, Taichung, Taiwan
    3. Department of Biotechnology, Asia University, Taichung, Taiwan
    Search for more papers by this author
  • Chung-Yi Wu,

    Corresponding author
    1. The Ph.D. Program for Cancer Biology and Drug Discovery, China Medical University, Taichung, Taiwan
    2. Genomics Research Center, Academia Sinica, Taipei, Taiwan
    • Correspondence to: Yung-Luen Yu, PhD, Graduate Institute of Cancer Biology, China Medical University, 9F, No. 6, Hsueh-Shih Road, Taichung 404, Taiwan, Tel.: +886-422052121 ext. 7933, Fax: +886-422333496, E-mail: ylyu@mail.cmu.edu.tw or Chung-Yi Wu, PhD, The Genomics Research Center, Academia Sinica, 128 Academia Road, Section 2, Nankang, Taipei 115, Taiwan, E-mail: cyiwu@gate.sinica.edu.tw

    Search for more papers by this author
  • Yung-Luen Yu

    Corresponding author
    1. The Ph.D. Program for Cancer Biology and Drug Discovery, China Medical University, Taichung, Taiwan
    2. Graduate Institute of Cancer Biology and Center for Molecular Medicine, China Medical University, Taichung, Taiwan
    3. Department of Biotechnology, Asia University, Taichung, Taiwan
    • Correspondence to: Yung-Luen Yu, PhD, Graduate Institute of Cancer Biology, China Medical University, 9F, No. 6, Hsueh-Shih Road, Taichung 404, Taiwan, Tel.: +886-422052121 ext. 7933, Fax: +886-422333496, E-mail: ylyu@mail.cmu.edu.tw or Chung-Yi Wu, PhD, The Genomics Research Center, Academia Sinica, 128 Academia Road, Section 2, Nankang, Taipei 115, Taiwan, E-mail: cyiwu@gate.sinica.edu.tw

    Search for more papers by this author

Abstract

Globo H, a cancer-associated carbohydrate antigen, is highly expressed in various types of cancers. However, the role of Globo H in hepatocellular carcinoma (HCC) remains elusive. In our study, we performed glycan microarray analysis of 134 human serum samples to explore anti-Globo H antibody changes and found that Globo H is upregulated in hepatitis B virus (HBV)-positive HCC. Similarly, immunohistochemistry showed that Globo H expression was higher in tumors compared to normal tissues. In addition, fucosyltransferase 2 (FUT2), the main synthetic enzyme of Globo H, was also increased in HCC cells overexpressing HBV X protein (HBX). HBX plays an important role in promoting cell proliferation and may be related to increased levels of FUT2 and Globo H. Furthermore, using microRNA profiling, we observed that microRNA-15b (miR-15b) was downregulated in patients with HCC and confirmed association of FUT2 expression with expression of its product, Globo H. Therefore, our results suggest that HBX suppressed the expression of miR-15b, which directly targeted FUT2 and then increased levels of Globo H to enhance HCC cell proliferation. Additionally, proliferation of HBX-overexpressing HCC cells was significantly inhibited by treatment with Globo H antibody in vitro. In xenograft animal experiments, we found that overexpression of miR-15b effectively suppressed tumor growth. The newly identified HBX/miR-15b/FUT2/Globo H axis suggests one possible molecular mechanism of HCC cell proliferation and represents a new potential therapeutic target for HCC treatment.

Abbreviations
CHB

chronic hepatitis B

FUT2

fucosyltransferase 2

HBV

hepatitis B virus

HBX

HBV X protein

HCC

hepatocellular carcinoma

HCV

hepatitis C virus

miR-15b

microRNA-15b

RT-PCR

reverse transcriptase-polymerase chain reaction

shRNA

short hairpin RNA

UTR

untranslated region

Hepatocellular carcinoma (HCC) is one of the most prevalent malignancies and a leading cause of cancer-related death worldwide, especially in East Asia and South Africa.[1] Hepatitis B virus (HBV) X protein (HBX), which is encoded by the HBV genome, is thought to play a key role in the molecular pathogenesis of HBV-related HCC.[2] HBX is critical for HCC development because it confers a proliferation privilege to cells, thus leading to malignancy.[2] Although HBX plays an important role in hepatocarcinogenesis, the underlying molecular mechanisms remain unknown.

MicroRNAs are evolutionarily conserved noncoding RNAs that are 21–25 nucleotides in length; they play critical roles in regulating gene expression and in multiple cellular processes including proliferation, development, differentiation and even tumorigenesis.[3] Alterations in miRNA expression are observed in HCCs and have been linked to the molecular pathogenesis of HCC, as they affect the expression of crucial mRNAs.[4] Depending on the target gene, miRNAs can function as tumor suppressor genes or oncogenes.

Globo H is a member of the family of antigenic carbohydrates that are highly expressed in various types of cancers, especially cancers of the breast, prostate, stomach and lung as cancer-associated carbohydrate antigens.[5] Globo H is a glycosphingolipid hexasaccharide with the structure Fucα(1-2)Galβ(1-3)GalNAcβ(1-3)Galα(1-4)Galβ(1-4)Glcβ(1).[6] Globo H is expressed on the cancer cell surface as a glycolipid and possibly as a glycoprotein,[7] but this has not been studied in HCC. Matta and coworkers demonstrated that Globo H is generated by adding fucose to the terminal galactose, and catalysis occurs in fractions containing α1,2-fucosyltransferase (FUT) activity.[8] Whether FUT1, FUT2 or both enzymes are involved in the synthesis of Globo H in HCC and whether they are affected by HBX remain unclear.

In our study, we first addressed the effect of HBX on Globo H regulation. Globo H expression was increased in HBV-related HCC, and HBX played an important role in this process. HBX activated cell proliferation, at least in part by regulating FUT2, which was inhibited by miR-15b. Finally, we validated the role of inhibition by anti-Globo H on hepatocarcinoma cell proliferation.

Material and Methods

Patient characteristics

Serum samples were collected at the National Cheng Kung University Hospital in Tainan, Taiwan. Samples were encrypted to protect patient confidentiality and were used under a protocol approved by the Institutional Review Board of Human Subjects Research Ethics Committee of Cheng Kung University Hospital (assignment number: ER-99-176). A total of 134 participants including 50 patients with HBV-related HCC, 61 patients with chronic hepatitis B (CHB) and 23 normal controls were recruited into our study (Supporting Information Table S1).

Microarray analysis

Globo H and its truncated analogs were synthesized as in Ref. [9]. The general procedure for fabrication of the glycan microarray was as previously described.[10] The mirVANA miRNA Bioarray V9.2 was purchased from Ambion (Austin, Texas).

Immunohistochemistry

Twenty-eight surgical pathology specimens [16 HBV-related and 12 hepatitis C virus (HCV)-related HCCs] were obtained from the HCC group participating in the serum study and used for the immunohistochemistry (IHC) assays. The IHC study for Globo H was performed on formalin-fixed, paraffin-embedded sections using the Bond polymer refine detection kit (Leica Microsystems, Wetzlar, Germany). The primary antibody used was a monoclonal murine anti-Globo H antibody (clone MBr1; 1:100; Alexis Biochemicals, San Diego, CA). The sections were deparaffinized in xylene and graded alcohols, hydrated, washed in phosphate-buffered saline (PBS) and immersed in 3% hydrogen peroxide for 20 min to block endogenous peroxidase activity. Sections were incubated with primary antibodies at room temperature overnight, followed by incubation with polyhorseradish peroxidase-conjugated anti-rabbit IgG. Diaminobenzidine was used to visualize the immunocomplex. Then, the sections were counterstained with hematoxylin, dehydrated, cleared and mounted. Processed slides were examined under a Nikon eclipse 80i microscope (Nikon, Tokyo, Japan).

Cell culture

HepG2, HepG2X, Hep3B and Hep3BX cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM)-F12 medium supplemented with 10% fetal bovine serum (HyClone, Logan, UT). HepG2X and Hep3BX cells were derivatives of human hepatoma Hep3B and HepG2 cells, respectively, stably expressing the HBX gene, and were established as described previously.[11] THLE-2 cells were maintained in complete growth medium BEGM (BEGM Bullet Kit; CC3170) (Lonza/Clonetics Corporation, Dublin, Ireland), and PLC/PRF/5 cells were cultured in DMEM (HyClone) supplemented with 10% fetal bovine serum. The THLE-2 and PLC/PRF/5 cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA) and Food Industry Research and Development Institute (FIRDI, Hsinchu, Taiwan), respectively. Cells were maintained in an atmosphere of 5% CO2 in a humidified 37°C incubator.

Plasmids, miRNA mimic, miRNA inhibitor and siRNA

The pcDNA6.0-HBX plasmid was constructed by cloning the cDNA product of the HBX gene into the pcDNA6.0 expression vector (Invitrogen, Grand Island, NY). The pCMV6-FUT2 plasmid was purchased from OriGene. The miR-15b mimic, miRNA mimic control, miRNA-15b inhibitor and miRNA inhibitor control were purchased from Dharmacon (Lafayette, CO). The microRNA mimic is a double-stranded oligonucleotide designed to mimic the function of endogenous mature miRNA. miRNA inhibitors are single-stranded, chemically enhanced RNA oligonucleotides designed to bind and sequester the complimentary, mature microRNA strand. The siRNAs that specifically target HBX and the siRNA control were obtained from Sigma (Saint Louis, MO). The siRNA sequences are shown in Supporting Information Table S2.

Transient transfection

Transfections were performed using Lipofectamine 2000 or RNAiMAX (Invitrogen), according to the manufacturer's instructions. After 48 hr, cells were used for RNA isolation, total lysate preparation, flow cytometry and proliferation assays.

Reverse transcriptase–polymerase chain reaction

Total RNA was extracted with TRIzol reagent (Invitrogen) according to the manufacturer's instructions. The primers were synthesized by Invitrogen (for sequences, see Supporting Information Table S2). Quantitative polymerase chain reaction (qPCR) was performed by detecting the hydrolyzed fluorescent probes from the Universal ProbeLibrary (Roche, Pleasanton, CA) using the LightCycler 480 apparatus (Roche).

Fluorescence cytometry analysis of Globo H antigens

Cells (1 × 106) were incubated with a monoclonal antibody against Globo H for 30 min at 4°C. After two washes, the cells were incubated in FITC-conjugated goat antibody against mouse IgM for 45 min at 4°C. The cells were then subjected to flow cytometry after washing. Data are expressed as the relative fluorescence intensity.

Gene knockdown with short hairpin RNA

Knockdown of genes was performed with specific short hairpin RNAs (shRNAs) delivered with the lentiviral system from the National RNAi Core Facility (Academia Sinica, Taipei, Taiwan) according to the instruction manual. The shRNA constructs targeting FUT2 are clones TRCN0000036102 and TRCN0000036103. The shRNA construct against luciferase (shLuc), clone TRCN0000072244, was used as a negative control. The efficiency of mRNA downregulation and the specificity for each sequence were tested with qPCR.

Dual-luciferase reporter assay

The sequence of the 3′-untranslated region (UTR) of FUT2 that is predicted to interact with miR-15b or a mutated sequence of this region was inserted into the SpeI and HindIII sites of the pMIR-REPORT Luciferase vector (Invitrogen) downstream of the luciferase gene. The primers were synthesized by Invitrogen (for sequences, see Supporting Information Table S2). Cells were harvested 48 hr after transfection and assayed with the Dual-Luciferase Reporter Assay (Promega, Fitchburg, WI). The pRL-TK (Promega) plasmid was used as an internal control. The results were expressed as relative luciferase activity (LUC/Renilla LUC). Each transfection was repeated thrice.

Western blot analysis

Cells were lysed in NETN buffer. Thirty micrograms of samples were run on 10% SDS-polyacrylamide gels, and the separated proteins were then blotted onto PVDF hybridization transfer membrane (Perkin-Elmer, Norwalk, CT). The primary antibodies used were mouse monoclonal anti-Hep B xAg (Santa Cruz Biotechnology, Dallas, Texas; 1:1,000), anti-FUT2 (Abcam, San Francisco, CA; 1:1,000), anti-α-tubulin (Sigma; 1:5,000) and anti-β-actin (Sigma; 1:5,000). The secondary antibodies used were horseradish peroxidase-conjugated rat anti-mouse secondary antibody (Chemicon, Billerica, MA; 1:5,000). Immunoreactive bands were visualized with the enhanced chemiluminescence detection reagent (GE, Piscataway, NJ).

Cell proliferation assay

Hep3B cells (5 × 103/mL) were cultured in 24-well plates for 24 hr and then transfected with different amounts of pcDNA6A-HBX plasmids. Control cells were treated with vector-only pcDNA6A. The number of viable cells was counted according to trypan blue dye exclusion[12] using a Countess Automated Cell Counter (Invitrogen). The monoclonal anti-Globo H VK-9 was a gift from Philip Livingston, Memorial Sloan-Kettering Cancer Center, New York.[13] In addition, to determine cell survival, cultures were treated with WST-1 (Roche) for 3 hr, and optical density was read on a spectrophotometer (BioTek) at 450 nm.

Tumor xenograft animal model

Four-week-old female athymic nude (nu/nu) mice were obtained from the National Laboratory Animal Center at Taiwan and used for subcutaneous tumor implantation. The miR-15b expression vector (Cat No. PMIRH15b-onlyAA-1) and empty vector (Cat No. PMIRH000AA-1) were purchased from System Biosciences (SBI, Mountain View, CA). To generate a stable cell line, PLC/PRF/5 cells were transfected with the vector using the Lipofectamine transfection method and were selected with 2 μg/mL puromycin for 2 weeks. PLC/PRF/5, PLC/PRF/5-miR-control or PLC/PRF/5-miR-15b cells (1 × 107 in 100 μL of sterile Dulbecco's PBS) were inoculated into nude mice by subcutaneous injection into the flank. Measurements of tumor size began 8 days postinoculation and continued every 5 days for 40 days. Tumor measurements were evaluated using calipers; tumor size was calculated using the following formula: (length × width2)/2.

Statistical analysis

Data are presented as the mean ± standard deviation. Data analysis was done using Prism 5 (GraphPad), Excel (Microsoft) and SigmaPlot 9.0 software (SigmaPlot). Statistical analyses were done with Student's t-test or Fisher's exact test.

Results

Globo H expression is increased in HBV-related HCC specimens

The serum of breast cancer patients contains high levels of antibodies against the Globo H epitope.[13] Thus, we evaluated the role of Globo H as a biomarker for HCC, especially HBV-related HCC. Using our highly sensitive Globo H microarray,[9] we examined serum samples from patients with HCC, CHB and normal controls for antibodies that bound to Globo H and its truncated fragments on glycan microarrays (Fig. 1a). The level of antibodies that bound to Gb5 was higher than the level of antibodies bound to other glycans in all samples, but the level of antibodies bound to Globo H was lower in CHB and normal donors compared to patients with HBV-related HCC (p = 0.001, 0.027, respectively) (Figs. 1b–1e). The levels of antibodies bound to Globo H analogs were much lower than Gb5 in HCC, CHB or healthy donors. However, the antibody levels for Bb2, Bb3 and Bb4 displayed no significant differences among these three groups. We also found no significant differences in antibody levels between HCV-related HCC patients and patients with chronic hepatitis C infection or normal donors (Supporting Information Fig. S1). Furthermore, Globo H was present in tissues, as detected with IHC (Fig. 1f). Immunoreactivity was observed in a subset of paired samples (adjacent nontumor and tumor tissue from the same patient). Two representative HBX-positive HCC specimens (case 1 and case 2) showed expression patterns of Globo H in tumor cells. All nontumor tissues were Globo H negative. Among the 16 HBV-associated cases, 12 tumor tissues (75%) showed Globo H-positive expression. In addition, only two cases (17%) were Globo H positive among the 12 HCV-associated cases. Fisher's exact test yielded p = 0.003, indicating a significant difference in the Globo H-positive/negative ratio between HBV-related and HCV-related HCC cases (Supporting Information Table S3). IHC examination showed that Globo H expression was also related to HBV-related HCC. Therefore, we decided to examine Globo H expression using microarray and IHC analysis in the same patient. The expression ratio of Globo H was significantly increased in patients with Globo H-positive staining compared to patients with Globo H-negative staining (p = 0.0002) (Fig. 1g). We hypothesize that the increased expression of Globo H and HBV infection are related.

Figure 1.

Ratios of IgG levels to Globo H analogs in sera from healthy individuals, patients with chronic hepatitis B (CHB) infection and those with HBV-related HCC. (a) A representative image obtained from a fluorescence scan after detection of IgG in sera samples. The grid contains eight glycans (Gb5, Gb4, Gb3, Gb2, Globo H, Bb4, Bb3 and Bb2) printed at a 100-μM concentration. The quantification data are shown below the images. These averages are for the five replicates shown on each array. The relative fluorescence ratios were obtained from the fluorescence intensity of Globo H (b) or Globo H analogs Bb2 (c), Bb3 (d) and Bb4 (e) divided by the fluorescence intensity of Gb5. The mean Globo H/Gb5 IgG ratio was significantly higher in sera from HCC patients. Box and whisker plots show the mean (line through the middle of the box), the 75th and 25th percentiles (the top and bottom of the box, respectively) and the 95th and 5th percentiles (the top and bottom of the whiskers, respectively). The p values were calculated with the Student's t-test (*p < 0.05). (f) Expression patterns of Globo H in nontumor and tumor tissues were assessed with IHC (lower panel; magnification, ×200 and inset, ×400). (g) A comparison of microarray results and IHC in specimens with negative and positive Globo H expression (data are the mean ± 95% confidence interval) (***p < 0.001).

Upregulation of Globo H expression by HBX in vitro and in vivo

To assess whether expression of Globo H is responsive to HBX overexpression, we first examined the HBX stable transfectants HepG2X and Hep3BX for expression of HBX compared to the parental cell lines HepG2 and Hep3B, respectively (Fig. 2a). Globo H levels were enhanced in HepG2X and Hep3BX cells compared to the parental cells (Fig. 2b). In addition, we introduced a plasmid encoding HBX into Hep3B cells and examined the expression of HBX (Fig. 2c). The expression of Globo H was significantly elevated in HBX-expressing 3B cells compared to empty vector-expressing Hep3B cells (Fig. 2d). To determine the relationship between HBX and Globo H, we analyzed the expression of HBX in 20 pairs of HCC tumor tissues with HBV and their corresponding adjacent nontumor liver tissue (T/NT ratio). We found a significant correlation between HBX and Globo H expression in HBV-related HCC tissues (r = 0.4559, p = 0.0434; Fig. 2e).

Figure 2.

Upregulation of Globo H expression by HBX. (a) HBX was abundantly expressed in HepG2X and Hep3BX cells. (b) Globo H expression as determined with flow cytometry. (c) HBX expression in Hep3B cells transiently transfected with a control vector plasmid or HBX plasmid. (d) Globo H expression in these transiently transfected cells was analyzed with flow cytometry. The p values were calculated with the Student's t-test (*p < 0.05). (e) Globo H expression levels were positively correlated with HBX expression levels in patients with HBV-related HCC. HBX gene expression levels are presented as log2 fold change values (T/NT). Correlation analyses were performed with linear regression models, and r and p values are indicated.

FUT2 is involved in the biosynthesis of Globo H in hepatocarcinoma cell lines

We next investigated the cellular and molecular effects of HBX upregulation in HCC cells. We thus investigated the expression of FUT1 and FUT2 in HepG2X and Hep3BX cells compared to their expression in parental cell lines. FUT1 mRNA expression was not different in the presence or absence of HBX, whereas FUT2 expression was significantly higher in HepG2X and Hep3BX cells compared to HepG2 and Hep3B cells, respectively (Figs. 3a and 3b). In addition, FUT2 mRNA was elevated along with the doses of HBX in 3B cells (Fig. 3c). Thus, FUT2 in particular may contribute to the increase in Globo H expression in the presence of HBX.

Figure 3.

FUT2 is involved in Globo H biosynthesis in hepatocarcinoma cell lines. (a) Differential expression of FUT1 and FUT2 mRNA in HepG2, HepG2X, Hep3B and Hep3BX cells. The expression of FUT1 and FUT2 was determined with qRT-PCR. (b) Western blot analysis of FUT2 protein. (c) Hep3B cells were transiently transfected with a control vector plasmid and/or HBX plasmid. FUT2 expression was analyzed with qRT-PCR (**p < 0.01). (d) The effect of shFUT2 in Hep3BX cells was evaluated with qRT-PCR. Total RNA was extracted 72 hr after infection with a control (shLuc) or shFUT2-encoding vector (shFUT2-1 and shFUT2-2). FUT2 mRNA expression is shown relative to the control. (e) The effect of shFUT2 in Hep3BX cells was evaluated with Western blotting. (f) Silencing of FUT2 mRNA in Hep3BX cells reduced Globo H expression.

To detect whether FUT2 was involved in Globo H expression, we downregulated FUT2 expression with shRNA that targets FUT2. A shLuc construct was included as a control. FUT2 mRNA and protein levels were greatly reduced in cells transfected with shRNA for FUT2 for 72 hr (Figs. 3d and 3e). The effect of downregulating FUT2 on the expression of Globo H was then investigated with flow cytometry analysis. Depletion of FUT2 for 72 hr led to decreased Globo H expression (Fig. 3f).

HBX suppresses miR-15b expression

To investigate the impact of HBX on miRNA expression, we hypothesized that any miRNA that plays a key regulatory role in HBX-induced FUT2 expression should also show an altered expression pattern in human hepatoma tissues. We characterized miRNA expression in HCC tissues harvested from two patients to compare HBV-positive HCC tumors with nontumor tissues. Twenty-six miRNAs were differentially regulated (≥2-fold change) in HCC tissues as shown by the T/NT ratios of these miRNAs (Supporting Information Table S4). We then used TargetScan to screen for putative miRNAs that target the 3′-UTR of FUT2. Of the 26 miRNAs that were differentially regulated, only miR-15b was associated with FUT2 (Fig. 4a). The results obtained for HBV-related HCC tissue samples identified FUT2 as a cellular target of miR-15b, with a mean T/NT ratio of 0.49 from the miRNA array analysis and 0.20 from quantitative reverse transcriptase (qRT)-PCR (Figs. 4b and 4c). To identify whether differentially regulated miRNA-15b was specifically associated with HBX in hepatocarcinoma cells, we tested cells that were stably or transiently transfected with HBX. The relative expression levels were determined with qRT-PCR. miR-15b expression was downregulated in HepG2X and Hep3BX cells compared to the parental cells (Fig. 4d). In addition, the level of miR-15b decreased according to the amount of HBX transfected into Hep3B cells (Fig. 4e). These data support the idea that HBX downregulates the expression of miR-15b. A correlation among miR-15b, FUT2 and Globo H expression was also confirmed in clinical samples. miR-15b expression levels were significantly inversely correlated with Globo H (r = −0.497, p = 0.03; Fig. 4f) and FUT2 (r = −0.545, p = 0.01; Fig. 4g) expression in 19 samples of human HCC tumor tissue. FUT2 was significantly correlated with Globo H (r = 0.499, p = 0.02; Fig. 4h).

Figure 4.

Validation of microarray data using qRT-PCR. (a) Intersection of analysis of the two profiles is shown for the miRNAs that are predicted to target FUT2 (TargetScan) and those that were differentially regulated. (b) miR-15b levels were downregulated in the tissues of two patients with HCC as determined with miRNA microarray analysis. Data are the fold changes in miRNA levels in tumor tissue relative to adjacent nontumor tissue, which was set as 1. (c) qRT-PCR analysis of miR-15b was performed to validate the microarray results. The relative amount of each miRNA was normalized to RNU6B. (d) miR-15b expression was determined in HepG2, HepG2X, Hep3B and Hep3BX cells. (e) miRNA-15b expression in Hep3B cells transiently transfected with the HBX plasmid. (f) miR-15b was negatively correlated with Globo H in 19 HCC samples. (g) miR-15b and FUT2 expression levels were inversely correlated in 19 HCC samples. (h) FUT2 was correlated with Globo H in 19 HCC samples. Correlation analyses were performed using linear regression models, and r and p values are indicated (*p < 0.05, **p < 0.01, ***p < 0.001).

FUT2 is a novel target of miR-15b

We tested whether miR-15b could downregulate FUT2. miR-15b showed only one conserved site (seed sequence matched with nt 631–633 and 640–646; Fig. 5a). We increased miR-15b expression by transfecting an miR-15b mimic into Hep3BX cells and using a mimic control as the negative control. A significant decrease in FUT2 mRNA levels was detected after upregulation of miR-15b (Fig. 5b). FUT2 protein was also decreased after transfection of the miR-15b mimic (Fig. 5c). To determine whether miR-15b directly recognizes the 3′-UTR of FUT2 mRNA, we cloned the 3′-UTR sequence containing the predicted target sites and a mutated sequence of the 3′-UTR downstream of the pMIR-REPORT™ Luciferase miRNA Expression Reporter Vector to generate pMIR-FUT2 and pMIR-FUT2-mut vectors, respectively. A Renilla luciferase vector was used for normalization of differences in transfection efficiency. Luciferase activity in Hep3BX cells cotransfected with the miR-15b mimic and the pMIR-FUT2 vector was significantly decreased compared to the mimic control (Fig. 5d). Thus, the 3′-UTR of the FUT2 transcript is a miR-15b target. Next, we determined whether miR-15b downregulates Globo H. The miR-15b mimics can be used to test downregulation of Globo H. Downregulation of Globo H was reversed by treatment with the FUT2 expression vector (Fig. 5e). Furthermore, similar results were obtained with HBX-transfected THLE-2 cells, which are immortalized human liver cells. The data suggest that HBX suppressed the expression of miR-15b and elevated the expressions of FUT2 and Globo H (Fig. 5f). We also employed RNAi to knock down the expression of HBX in an HBV-expressing HCC cell line, PLC/PRF/5, an HCC cell line that expresses HBX naturally. We also demonstrated that knocking down the expression of HBX increased miR-15b and decreased both FUT2 and Globo H (Fig. 5g).

Figure 5.

FUT2 is a target of miR-15b. (a) Binding sites of miR-15b predicted in the FUT2 3′-UTR by TargetScan. (b) The relative FUT2 mRNA levels in Hep3BX cells 48 hr after transfection of the miR-15b mimic (40 and 60 nM) or miRNA mimic control. β-Actin was used as an endogenous control. (c) Immunodetection of FUT2 in whole-cell lysates 48 hr after transfection of the miR-15b mimic (40 and 60 nM) or miRNA mimic control into Hep3BX cells and nontransfected Hep3BX cells. α-Tubulin was used as an endogenous control. (d) Effect of miR-15b on expression of FUT2 with the luciferase reporter assay. Luciferase activity was determined after cotransfection of pMIR-FUT2 or pMIR-FUT2-mut with 60 nM miR-15b mimic or mimic control. (e) Flow cytometry analysis of Globo H in Hep3BX cells transfected with miR-15b, an appropriate control and/or the FUT2 expression plasmid. Immunoblot under the histograms shows corresponding FUT2 expression. (f) THLE-2 cells transfected with HBX or empty vector and (g) PLC/PRF/5 cells transfected with HBX siRNA or control siRNAs were assessed for FUT2, miR-15b and Globo H levels (*p < 0.05, **p < 0.01, ***p < 0.001).

Functional effects of HBX/miR-15b/FUT2/Globo H on tumor cell proliferation

To determine whether the increase in cell proliferation was mediated by HBX, we compared the growth rates of cells transduced with HBX or an empty vector control. A trypan blue exclusion assay indicated that HBX promoted HepG2 and Hep3B cell proliferation in vitro (Fig. 6a). To test the effect on cell proliferation after upregulation or downregulation of miR-15b, transfection of the miR-15b mimic and inhibitor was performed. The miR-15b mimic decreased cell proliferation, whereas the miR-15b inhibitor enhanced cell proliferation as shown by the WST-1 assay (Figs. 6b and 6c). These observations collectively suggest that miR-15b overexpression in HCC cells may reduce cell proliferation. We next asked if the effects of miR-15b on cell growth are related to FUT2 expression. We knocked down FUT2 levels in HepG2X and Hep3BX cells with the two shRNA FUT2 sequences, resulting in a decrease in growth compared to cells treated with control (shLuc) (Fig. 6d). Finally, we determined if anti-Globo H (VK-9 and Mbr-1) affected cell proliferation in HepG2X and Hep3BX cells compared to nonspecific antibodies (IgG and IgM). VK-9 and Mbr-1 significantly inhibited HepG2X and Hep3BX cell proliferation as assessed by the WST-1 assay (Figs. 6e and 6f). To further identify the function of miR-15b on inhibition of tumor growth in vivo, PLC/PRF/5, PLC/PRF/5-miR-control or PLC/PRF/5-miR-15b cells were subcutaneously inoculated into mice (n = 4). Although tumors formed in flank inoculated with either PLC/PRF/5-miR-control or PLC/PRF/5-miR-15b cells, tumor growth of the PLC/PRF/5-miR-15b-induced tumors was significantly reduced (Fig. 6g). The mean tumor volume of the PLC/PRF/5-miR-control group was 8.9-fold higher than that of the miR-15b group (600 ± 157 versus 67 ± 9 mm3, p < 0.01) at 40 days postinoculation.

Figure 6.

Determination of the effect of HBX/miR-15b/FUT2/Globo H on hepatoma cell growth. (a) Growth curves were determined in HepG2 and Hep3B cells. HepG2 and Hep3B cells were transiently transfected with a control vector plasmid and/or HBX plasmid. HBX induced hepatoma cell proliferation. (b, c) Effects of miR-15b overexpression or downregulation on HepG2X and Hep3BX cell growth. HepG2X and Hep3BX cells were transfected with 60 nM miR-15b mimic, mimic control, miR-15b inhibitor or inhibitor control, and cell growth was determined with a WST-1 assay after transfection for 48 hr. (d) HepG2X and Hep3BX cells were treated with shLuc or shFUT2, and growth curves were determined. (e, f) Effect of anti-Globo H VK-9 and MBr-1 on the proliferation of HepG2X and Hep3BX cells. (g) Tumorigenesis in a subcutaneous xenograft of stable subclones. The tumor growth curve was measured every 5 days for 40 days after inoculation (n = 4). The mean tumor volume was significantly reduced by miR-15b overexpression (means ± SE). The tumor mass was harvested from sacrificed mice on day 40 after inoculation. Calipers were used to determine the tumor mass. The p values were calculated with the Student's t-test (*p < 0.05, **p < 0.01, ***p < 0.001).

FUT2 expression is associated with tumor size

We next asked if FUT2 expression was affected during liver cancer progression. In a subset of 24 paired samples (tumor tissue compared to nontumor tissue from the same patient; T/NT ratio), we found a statistically significant decrease in FUT2 expression in tumors ≤3 cm in diameter compared to tumors >3 cm (Supporting Information Fig. S2a). However, no statistically significant upregulation of FUT2 was observed in relation to vascular invasion, pathological staging or clinical staging (Supporting Information Figs. S2b–S2d). Therefore, our data suggest that FUT2 increases hepatocyte growth, which is essential for HCC progression or recurrence.

Discussion

The HBX protein is a multifunctional protein that not only activates transcriptional transactivation but also mediates cell growth via proliferation and apoptosis.[14] Our report and another study[15] have demonstrated that HBX promotes proliferation of HCC cells. HBX alters the in vitro expression of multiple miRNAs in malignant hepatocytes, particularly those of the miR-16 family.[16] The miR-16 family is composed of miR-15a, -15b and -16. The miR-15a/16-1 and miR-15b/16-2 gene clusters are located on human chromosomes 13q and 3, respectively.[17] This family is repressed and suppresses (in specific tumor types) a series of oncogenes that include BCL2, cyclin D1 and WNT3A.[18] However, among miR-16 family members, only miR-15b was suppressed in our miRNA array analysis. The disparities between these data may have resulted from differences in the clinical tissue samples and in microarray sensitivity. Recent studies have identified that the HBX transcript directly triggers downregulation of miR-15a/miR-16-1 via microRNA targeting sequences in the viral RNA.[19] Moreover, miR-15a and miR-15b share most of the same sequence. This finding provides one possible explanation of regulation of miR-15b by HBX. In addition, the functional interaction between miR-15b and its target gene may also account for the inhibition of vessel formation and its invasive effect.[20, 21] Thus, miR-15b could be a potential tumor suppressor in HCC.

Few studies have been published regarding associations between miRNAs and fucosyltransferases. Here, we found that HBX altered miR-15b expression and, by extension, downregulated FUT2. FUT1 is expressed on erythrocyte membranes and vascular endothelium,[22] but FUT2 is mainly expressed in the mucosal epithelium of buccal tissue, breast and body fluids as well as in the gastrointestinal, respiratory and genitourinary tracts.[23] HBX upregulated expression of FUT2, although FUT1 expression was considerably weaker. Silencing of FUT2 led to a decrease in Globo H antigen expression and cell proliferation. Overexpression and knockdown studies both revealed that FUT2 enhanced proliferation of HCC cells, and silencing FUT2 dramatically reduced cell growth. Recently, a study reported inhibition of cell growth by suppression of FUT1/2 expression in human epithelial carcinoma A431 cells.[24] Fucosylated carbohydrate structures are involved in many biological processes[25, 26] including tissue development, angiogenesis, fertilization, cell adhesion, inflammation and tumor metastasis. Here, we provide evidence of the relationship between HBX and FUT2 that is responsible for increased expression of Globo H. Therefore, investigation of the regulatory mechanism of proliferation that is mediated by Globo H is of interest. Both FUT1 and FUT2 mediate alpha-1,2 linkage of fucose to Gb5 to generate Globo H and the expression of FUT1 and FUT2 in various cancer cells has different levels. In our study, we showed that HBX upregulated expression of FUT2, but the Globo H-expressing breast cancer cell lines (MCF-7 and MB157) has been shown to highly express the FUT1.[27] Previous studies have demonstrated the activation of FUT1 and FUT2 by IL-1β inflammatory cytokinesis through the NF-κB pathway in gastric cancer cells.[28] However, what mechanism upregulates FUT1/Globo H or FUT2/Globo H in non-HBV-infected cancer cells or other human cancers have yet to be clearly delineated.

Our data demonstrate the clinical relevance of miR-15b, FUT2 and Globo H. Clinically, Chung et al. reported that high expression of miR-15b predicts a low risk of tumor recurrence after curative resection of HCC, suggesting its prognostic significance.[29] Certainly, small molecules will help improve the clinical management of HCC in the future. Aberrant glycosylation associated with tumor progression was first described in 1969 by Meezan et al.[30] who demonstrated that many glycans on cancer cells differ from glycans on normal cells. Globo H is overexpressed on a variety of tumor types. We first reported that IHC for Globo H was positive in 12 of 16 HBV-related HCCs (75%). Changes in glycosylation are associated with physiological and pathophysiological conditions of cells. Our previous data showed that the levels of IgG against Globo H are not significantly higher in patients with HCC than in healthy individuals.[10] However, the levels of antibodies bound to Globo H were much higher in HBV-infected patients with HCC than in those with CHB or healthy donors in our study. Taken together, these previous data and our results suggest that Globo H has potential value as a molecular target in cancer therapy.

The results of our study indicate that miR-15b overexpression in HCC cells may reduce tumor growth. Moreover, the results of a previous study also suggest that modulation of miR-15b expression may be therapeutically useful as an apoptosis-sensitizing strategy.[29] Thus, miR-15b may be a potential therapeutic or diagnostic/prognostic target for HCC. In addition, we demonstrated that treatment of liver cancer cells with anti-Globo H antibodies blocked HBX-induced cell proliferation. Several groups have performed or are in the process of carrying out clinical trials with monovalent and polyvalent vaccines against Globo H because of its attractiveness as a therapeutic target.[31-35] Our finding that Globo H antibodies can suppress the growth of HCC cell lines is consistent with the idea behind these studies. All these findings support a rationale for developing carbohydrate-based vaccines based on Globo H. Finally, the glycan microarray offers a powerful platform for testing the specificity of antibodies.

In summary, our results provide new insights (Supporting Information Fig. S3) into the miR-15b/FUT2/Globo H pathway that promotes HCC proliferation induced by HBX. Thus, miR-15b and Globo H represent key diagnostic markers and potential therapeutic targets for the treatment of HBV-related HCC.

Acknowledgements

The authors thank the National RNAi Core Facility (Academia Sinica, Taipei, Taiwan) for providing the shRNAs. They also thank Dr. Timothy C. Taylor (BIOMEDITOR, International Bioscience Consultants) for editing this manuscript.

Ancillary