The expression level of miR-18b in hepatocellular carcinoma is associated with the grade of malignancy and prognosis

110 hepatocellular carcinoma tissue samples were obtained by surgical resection (Additional file 1: Table S1). All patients or their guardians provided written informed consent, and Osaka City University, Ogaki Municipal Hospital and Kyoto University Graduate School and Faculty of Medicine’s Ethics Committee approved all aspects of this study in accordance with the Helsinki Declaration.

Total RNA from cell lines or tissue samples was prepared using a mirVana miRNA extraction Kit (Ambion, Austin, TX, USA) according to the manufacturer’s instruction. To detect miRNA, 100 ng of RNA was labeled and hybridized using the Human microRNA Microarray Kit (Rel 12.0) (Agilent Technologies, CA, USA) according to manufacturer’s protocol for use with Agilent microRNA microarrays Version 1.0. Hybridization signals were detected with Agilent DNA microarray scanner G2505B and the scanned images were analyzed using Agilent feature extraction software (v9.5.3.1). Data were analyzed using GeneSpring GX 7.3.1 software (Agilent Technologies) and normalized as follows: (i) Values below 0.01 were set to 0.01. (ii) Each measurement was divided by the 75th percentile of all measurements from the same species. All data were deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE31164.

To detect miRNA level by real-time qPCR, TaqMan® microRNA assay (Applied Biosystems) was used to quantify the relative expression level of miR-18b (assay ID. 002217); U18 (assay ID. 001204) was used as an internal control. cDNA was synthesized using the Taqman miRNA RT Kit (Applied Biosystems). Total RNA (10 ng/ml) in 5ml of nuclease free water was added to 3 ml of 5× RT primer, 10× 1.5 μl of reverse transcriptase buffer, 0.15 μl of 100 mM dNTP, 0.19 μl of RNase inhibitor, 4.16 μl of nuclease free water, and 50U of reverse transcriptase in a total volume of 15 μl. The reaction was performed in triplicate for 30 min at 16°C, 30 min at 42°C, and 5 min at 85°C. Chromo 4 detector (BIO-RAD) was used to detect miRNA expression.

The human hepatoma cell lines Huh-7, Li7 and human embryonal kidney cells lines 293FT were obtained from Japanese Collection of Research Bioresources cell bank. Cells were maintained in D-MEM (Invitrogen, Carlsbad, CA, USA) with 10% fetal bovine serum and plated in 6-well plates the day before transfection, then grown to 70% confluence. Cells were transfected with 12.5 pmol/l of Silencer® negative control siRNA (Ambion), siTNRC6B s; 5′-gggacaaggaggaaagaaatt-3′, as; 5′-uuucuuuccuccuuguccctt-3′ (Hokkaido System Science, Sapporo, Japan) or double-stranded mature miR-18b (Hokkaido System Science) using lipofectamine RNAiMAX (Invitrogen). Cells were also transfected with 1 μg/μl of negative control cDNA empty vector or total TNRC6B expression vector (Addgene) using Fugene® (Roche). TNRC6B complete plasmid set was obtained from the non-profit repository AddGene (http://​www.​addgene.​com). Cells were harvested 48 hr after transfection.

cDNA was synthesized using the Transcriptor High Fiderity cDNA synthesis Kit (Roche, Basel, Switzerland). Total RNA (2 μg) in 10.4 μl of nuclease free water was added to 1 μl of 50 mM random hexamer and denatured for 10 min at 65°C. The denatured RNA mixture was added to 4 μl of 5× reverse transcriptase buffer, 2 μl of 10 mM dNTP, 0.5 μl of 40U/μl RNase inhibitor, and 1.1 μl of reverse transcriptase (FastStart Universal SYBR Green Master Roche) in a total volume of 20 μl. The reaction was run in triplicate for 30 min at 50°C (cDNA synthesis), and five min at 85°C (enzyme denaturation). Chromo 4 detector (BIO-RAD, Hercules, CA, USA) was used to detect mRNA expression. The primer sequences was as follows TNRC6B s; 5′-acaagtgacaggagcgctgctg-3′, as; 5′- ccatgtcagacccgtctacaat-3′, and β-actin s; 5′-ccactggcatcgtgatggac-3′, as; 5′-tcattgccaatggtgatgacct-3′. Assays were performed in triplicate, and the expression levels of target genes were normalized to the expression of the β-actin gene (internal control), as quantified by real-time qPCR.

To generate the TNRC6B 3′-UTR luciferase reporter vectors, the TNRC6B 3′-UTR segments were generated (Additional file 1: Table S3) and inserted into the pMIR-REPORT Luciferase vector between SpeI and HindIII sites (Ambion). Wild and mutant type reporter vectors with miR-18b complementary sites were confirmed by sequencing. Huh7.5 cells (5×10⁴ cell/well) were transfected into 24-well dishes with DMRIE-C (Invitrogen), 10pmol of double stranded mature miR-18b, Antisense oligonucleotide of miR-18b, or negative control of RNA and 0.25 μg wild or mutant type reporter vector. After 48 h, the transfected cells were harvested and lysed, and their luciferase activity was measured with a Dual-Luciferase Reporter Assay System kit (Promega). The experiments were repeated at least three times.

A cell lysate from RNA-induced silencing complex (RISC), was collected using microRNA Isolation Kit, Human Ago2 (Wako, Osaka, Japan) according to the manufacturer’s instruction. Briefly, 48 hr after transfection with 12.5 pmol/L of double stranded mature miRNA, 5×10⁶ cells in 6 cm dish were washed with PBS and harvested by trypsinization. The cell pellet was re-suspended with the gentle pipetting in 1ml of cell lysis buffer (microRNA Isolation Kit). The cell suspension was incubated for 10 minutes on ice, and was then centrifuged at a force of 20000 g for 20 minutes at 4°C. The cells were suspended in PBS and mixed with beads conjugated human Argonaute2 (hAgo2) monoclonal antibody for 2 hr at 4°C. RNA from Ago2-immunoprecipitation fraction was then extracted using mirVana miRNA extraction kit.

Eight paraffin-embedded tissue samples (case 64, 108, 248, 261, 274, 277, 310 and 333) were used (Additional file 1: Table S1). We generated both a locked nucleic acid (LNA) − modified probe for miR-18b (5′- taaggtgcatctagtgcagttag-3′) and a scrambled negative control sequence (5′-gtgtaacacgtctatacgccca-3′: miRCURY-LNA detection probe, Exiqon, Vedbaek, Denmark). In situ hybridization utilized a RiboMap in situ hybridization kit (Roche Diagnostic) on a Ventana Discovery automated in situ hybridization instrument (Roche Diagnostic). For immunohistochemistry of FFPE sections, we used the Ventana HX System Benchmark (Roche Diagnostic). TNRC6B antibody (HPA003180) was purchased from Sigma-Aldrich, St. Louis, MO, USA.

Positive in situ hybridization staining and immunostaining were interpreted semi-quantitatively by assessing the intensity and extent of staining on the entire tissue sections observed on the slides, as described previously [18].

The cell proliferation assay was performed using XTT® Cell Proliferation Assay Kit (Roche). Briefly, huh7 and Li7 cells (5×105 cells/ml) were spread into 96-well dishes. 12.5 pmol/l of double stranded mature miR-18b, 2′-O-methylated antisense oligonucleotide (ASO) of miR-18b (Hokkaido System Science), siRNA for TNRC6B (Hokkaido System Science) and Silencer® negative control siRNA (Ambion), were transfected with lipofectamine RNAiMAX (Invitrogen). 2 ug of plasmids containing the TNRC6B (Addgene), or empty vector pcDNA3 (Invitrogen) was transfected with FuGENE 6 (Roche). After 24 or 72 hr of transfection, cells were washed twice with PBS, 50ul of XTT labeling mixture was added, and then cells were incubated in a humidified atmosphere for 6 hr at 37°C. After incubation, the absorbance of samples was measured using an ELISA reader at 450–500 nm against a reference wavelength of 650 nm.

The cell adhesion assay was carried out using Vybrant® Cell Adhesion Assay Kit (Invitrogen). Briefly, transfection procedure was same as the cell proliferation assays. After 24 or 72 hr of transfection, cells were washed twice with PBS then re-suspended in serum free D-MEM and incubated with 5 μl of the calcein AM stock solution at 37°C for 30 min. Following this, cells (5×10⁵/ml) were washed twice with D-MEM and re-suspended in D-MEM. The calcein-labeled cells were incubated at 37°C. After 120 min, non adherent cells were removed by washing and fluorescence was measured with a fluorescein filter set (absorbance 494 nm, emission 517 nm). We determined the percentage of adhesion by dividing the corrected (background subtracted) fluorescence of adherent cells by the total corrected fluorescence of cells added to each well.

Statistical analyses were performed using Student’s t-test; p values less than 0.05 were considered statistically significant. Microarray data were also statistically analyzed using ANOVA or Welch’s test and Bonferroni correction for multiple hypotheses testing. Survival analysis was used with Kaplan-Meier survival curve and log-rank tests in the R software environment.

miRNA expression profiles in 110 HCC were established by microarray analysis (Table 1 and Additional file 1: Table S1) and comprised of 60 moderately, 30 poorly, and 20 well differentiated HCC. We chose miRNAs that were clearly expressed in at least 70% of all samples as determined by numeric analysis. Twelve miRNAs were significantly differentially expressed depending on whether HCC was poorly, moderately or well differentiated according to ANOVA analysis. The expression of miR-221, miR-18a, miR-18b, and miR-423-5p in poorly differentiated HCC were significantly higher than in well differentiated HCC, and 8 miRNAs (miR-455-3p, miR-1914*, miR-100, miR-215, miR-122*, let-7b, miR-22 and miR-99a) in poorly differentiated HCC had significantly lower expression levels than in well differentiated HCC (p < 0.05) (Table 2).

We then detected the target gene of the 12 miRNAs that were differentially expressed according to the level of HCC differentiation. Homo sapiens trinucleotide repeat containing 6B (TNRC6B) was a common hypothetical target gene in miR-221, miR-18a, miR-18b, miR-423-5p, miR-455-3p, miR-1914*, miR-215, miR-122*, let-7b, and miR-22 using miRanda algorithm. TNRC6B on the other hand, was a common target gene in miR-221, miR-18a, miR-18b, miR-423-5p, and miR-22 using Targetscan. miR-221, miR-18a, miR-18b, miR-423-5p, and miR-22 could recognize TNRC6B as a target gene using both algorithms (Figure 1A)

To clarify the biological links between miRNAs and TNRC6B, we examined the expression pattern of TNRC6B in Huh7 cells by real-time qPCR when expression levels of miR-18a, miR-18b miR-122, miR-221, miR-423-5p, and miR-22 were either over-expressed or suppressed. The result was that low expression of TNRC6B was reflected by over-expression of miR-18b treated with mature miR-18b and vise versa when miR-18b was suppressed with antisense oligonucleotide. The expression pattern of TNRC6B when miR-22 was over-expressed or suppressed was similar to the expression pattern using miR-18b (Figure 1B). However, over-expression of miR-18a, miR-122, and miR-423-5p did not suppress the expression level of TNRC6B and suppression of miR-221 did not induce over-expression of TNRC6B.

Based on our findings, we prepared the reporter gene assay with wild or mutant sequence of the hypothetical binding site of TNRC6B and miR-18b. When miR-18b was co-transfected with wild type of 3′UTR of TNRC6B reporter genes, we observed that luciferase activity was significantly low compared to co-transfecting control RNA or miR-18b plus ASO miR-18b with wild type vector. However, in case of the mutant form of 3’UTR of TNRC6B reporter vector, the luciferase activity was not affected by transfection of any miRNAs (Figure 1C).

Then, we speculated that miR-18b and miR-22 could regulate the expression level of TNRC6B, and to clarify this physiological association, we performed an Ago2-coimmunoprecipitation (Ago2-IP) analysis. Ago2-IP fractionated cell lysates were prepared by transfecting 293FT cells with mature double strand of miR-18b, miR-22, miR-455-3p, let-7b, or a non-specific siRNA which was used as a control RNA. miR-455-3p and let-7b were used as negative controls. The TNRC6B RNA in the Ago2-IP fraction (IP RNA) was quantified by real-time qPCR. The concentration of TNRC6B IP-RNA treated with miR-18b was higher than those treated with the control RNA or double strand of miR-22, miR-455-3p, and let-7b (Figure 1D). Taken together, we concluded that miR-18b can regulate the expression of TNRC6B as a target gene.

We compared TNRC6B expression in clinical samples with the grade of histological differentiation. TNRC6B expression in poorly differentiated HCC was lower than in well differentiated HCC (Figure 2A). miR-18b confirmed the microarray results using real-time qPCR, while the real-time qPCR result corresponded to the microarray analysis result (Figure 2B). Although the microarray and realtime qPCR results correlated, the expression level of miRNA in both experiments differed. One reason for this difference could be that the base sequences of probe which recognizes miR-18b differ. However, there was no significant correlation between the contra-relation of miR-18b and TNRC6B expression. We then performed in situ hybridization using locked nucleic acid (LNA)-modified probes labeled with digoxigenin (DIG) in miR-18b and also immunohistochemistry of TNRC6B in 8 samples (well differentiated: case 64, 108; moderately differentiated: 248, 277, 310, 333; poorly differentiated: case 261, 274). We found high miR-18b and low TNRC6B expression levels in poorly differentiated HCC (case 261), low miR-18b and high TNRC6B expression levels in highly differentiated HCC (case 64), and moderate expression miR-18b and TNRC6B in moderately differentiated HCC (case 248) (Figure 3A, B). The results of the five remaining cases and scrambled LNA study are not shown.

Over-expression of miR-18b and inhibition of TNRC6B by siRNA in human hepatoma cell lines Huh7, showed the progression of cell proliferation. Inhibition of TNRC6B by siRNA in both Huh7 and Li7, showed the progression of cell proliferation (p < 0.05) (Figure 4A). In both cell lines, over-expression or inhibition of miR-18b showed deceleration or acceleration of cell adhesion respectively (p < 0.05). Over-expression of TNRC6B by siRNA showed acceleration of cell adhesion in Huh7 (p < 0.01); however, over–expression of TNRC6B by siRNA also showed a similar acceleration of cell adhesion in Li7 (Figure 4B). These results indicated that over-expression of miR-18b and inhibition of TNRC6B, have the advantage of accelerating cell proliferation and decelerating cell adhesion.

We then analyzed the prognosis of 73 HCC whose progress was monitored after surgery resection. Kaplan-Meier survival analysis and log-rank test demonstrated a significant difference in the outcomes of patients who were divided into two groups based on their median miR-18b expression level (p < 0.05). Specifically, patients with high expression of miR-18b had significantly lower survival rate than patients with low miR-18b expression. While miR-18b expression was associated with the relapse-free rate after surgical resection, we found that it did not significantly affect the overall survival rate (Figure 5 and Additional file 1: Table S2).

To ascertain if any connection exists between miRNA expression and clinical background, we compared miRNA expression with tumor size, gender, age and background of HCC. Since a 20 mm diameter tumor is standard for liver cancer in the early stage, we compared the miRNA expression for HCC larger than 20 mm with those smaller than 20 mm. Three miRNAs were extracted based on two criteria: fold change 0.5 > or 2.0<, and t-test p < 0.05. The expression level of miR-1471 in small HCC was significantly higher than in large HCC, and the expression level of miR-499-5p and miR-609 in small HCC was significantly lower than in large HCC (Table 3).

We also identified miRNAs with expression levels that varied according to gender and age. Namely, miR-765, miR-622, and miR-1300 had significantly lower expression levels in female HCC than in male HCC (Table 2). In regards to age, we discovered that the expression levels of 14 miRNAs (miR-654-5p, miR-493*, miR-410, miR-376a*, miR-758, miR-381, miR-543, miR-539, miR-487b, miR-337-5p, miR-136*, miR-154*, miR-330-3p, and miR-421) were significantly higher in HCC up to 66 years old than in HCC over 67 years old. The average age of the HCC subjects was 66.8 years old (Table 3).

Finally, when miRNA expression pattern was linked to the cause of HCC, we found that the expression level of miR-181d, miR-542-3p, and miR-519e in HCC derived from CH was significantly higher than in HCC from liver cirrhosis (LC). Additionally, the expression level of miR-939 in HCC derived from CH was significantly lower than in HCC from LC (Table 3). However, we found no significant correlation between the expression pattern of miR-18b and tumor size, age, gender, and background of HCC.