CADM2, as a new target of miR-10b, promotes tumor metastasis through FAK/AKT pathway in hepatocellular carcinoma

From January 2017 to October 2017, 36 fresh primary tumor samples and their corresponding, non-tumorous tissues were obtained from hepatic carcinoma patients at Heilongjiang Cancer Hospital (Harbin, China). Consents from each patient and approval by the local ethics committee were obtained (for the characteristics of all patients, see Additional file 1: Table S1). All specimens were histopathologically confirmed. HCC was graded according to the World Health Organization grading system and staged according to the American Joint Committee on Cancer (AJCC) tumor node-metastasis (TNM) staging system.

Human normal liver cell line HL7702 and HCC cell lines HepG2, Huh-7, Hep3B were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS, GIBCO, Carlsbad, CA, USA), 100 U/ml penicillin G and 100 μg/ml streptomycin at 37 °C in a humidified incubator containing 5% CO₂. LY294002 (#HY-10108, 2-morpholin-4-yl-8-phenylchromen-4-one) was purchased from MedChem Express (MCE, USA). 5 mg/ml LY294002 (storage solution) was prepared using Dimethyl Sulfoxide (DMSO).

MiR-10b mimics and miR-10b inhibitors as well as their corresponding negative control were purchased from GenePharma (Shanghai, China). HepG2,Huh7 and Hep3B cells were transfected with miR-10b mimic or miR-10b inhibitor using Lipofectamine 2000 transfection reagent (Invitrogen, Carlsbad, CA) according to reagent protocols. The oligonucleotide sequences utilized in transfection had been listed in Additional file 2: Table S3. Plasmid pEZ-M98-CADM2 (pEX-CADM2) and its corresponding empty vector (pEX-NC) were purchased from GeneCopoeia (GeneCopoeia, USA). To overexpress CADM2 in HCC cell lines, pEX-CADM2 or pEX-NC was transfected using the method described previously [22]. Co-transfection of miR-10b mimic and pEX-CADM2 was conducted using Lipofectamine 2000 Reagent (Invitrogen, Carlsbad, CA).

Cultured cells were harvested and total RNA extracted by TRIzol Reagent (Life Technologies, USA). Then reverse transcription operated with ABI High Capacity cDNA reverse transcription Kit (Thermo Fisher Scientific, USA) as described previously [22]. The primers utilized in qRT-PCR had been listed in Additional file 2: Table S3.

Total protein was extracted using RIPA buffer and protein expression was analyzed by Western blotting as described previously [22]. GAPDH served as an endogenous control. The antibodies information utilized in Western blotting had been listed in Additional file 3: Table S4.

Cells were rinsed with PBS and fixed with 4% paraformaldehyde for 30 min at room temperature followed by permeabilization with 0.1% sodium citrate plus 0.1% Triton X-100. The cells were subjected to immunofluorescent staining with primary antibody (Additional file 3: Table S4) for 16 h at 4 °C. Cells were then washed with cold PBS three times for 5 min each and incubated with fluorescence labeled secondary antibody (1:500, #ZF0511, ZSGB-BIO) for 30 min. The cells were visualized using inverted fluorescence microscope (FSX100, Olympus).

A Transwell system (Corning Life Sciences) containing a polycarbonate filter (6.5 mm in diameter, 8 μm pore size) was used for migration and invasion assay. For cell invasion assay, the membrane undersurface was coated with 50 μl of matrigel mixed with DMEM at a 1:8 dilution and subsequently applied to the topside of the filter. By contrast, the filter was not coated for the cell migration assay. In both assays, cell suspensions (2 × 10⁴cells/well) were added to the upper chamber in medium without serum. Medium containing 1% FBS in the lower chamber served as a chemo-attractant. The cells that did not migrate or invade after 24 h of incubation were removed from the upper face of the filters by scrubbing with a cotton swab. Membranes were then fixed with 4% formaldehyde for 30 min at room temperature and stained with 0.1% crystal violet for 15 min. Finally, the number of migrating or invading cells was counted at × 200 magnification from ten different fields for each filter and analyzed to determine statistically significance.

A wound-healing assay was also applied to evaluate the cell migration ability. Cells were seeded in 3.5-cm plates and grown to a density of 70–80%. Then, a 200 μl pipette tip was used to create an artificial wound of scratched cells. The migrating distance was measured after 48 h or 72 h.

To construct a pmiR-CADM2–3′UTR plasmid containing the potential miR-10b binding sites, an 1168-bp sequence was amplified and inserted into the SacI and XbaI sites of the pmir-GLO Dual Luciferase vector (Promega, Madison, WI, USA). This sequence contained the two predicted binding sites at 6767 nt–6775 nt and 7543 nt–7551 nt.

The plasmid with mutant-type (MUT1, the first binding site is mutated; MUT2, the second binding site is mutated) were inserted downstream of the luciferase of pmirGLO Dual-Luciferase vector. HEK293 cells were used to measure luciferase activity. When grown to 60–70% confluence, the cells were co-transfected with a 100 ng Luciferase plasmid along with a 60 pmol miR-10b mimic or NC mimic as described above. After incubation for 24 h at 37 °C, the luciferase activity was determined using the Dual Luciferase Reporter 1000 Assay System (Promega, Madison, WI, USA).

For mRNA expression analysis, the RNA-seq V2 data was obtained from TCGA (The Cancer Genome Atlas) database. The normalized RSEM value was extracted, log2 transformed and merged. Samples were grouped into cancerous and normal tissues based on barcodes. The mean, max, min and SD of each group were calculated and paired student’s t test was preformed to verify the statistical significance. For gene methylation analysis, the mean beat value was extracted from TCGA. Samples were matched base on barcodes. The Pearson’s correlation was calculated to test the link between CADM2 mRNA expression and methylation. The survival analysis was followed the method in Broad institute TCGA Genome Data Analysis Center. In brief, the normalized RSEM value and clinical data were obtained from TCGA database. Data was processed using R. Expression value of miR-10b was extracted and filtered. Samples that have low expression value (RSEM < 1) were excluded. The mean, median and SD were calculated. Patients were grouped based on miR-10b expression. High expressed group, in the group with high expression level, patients with a top 20% miR-10b expression level (n = 86). Low expressed group, patients with a low 20% miR-10b expression level (n = 87). The Kaplan-Meier plot was generated by “survival” package in R.

SPSS was used for the statistical analysis. All values are expressed as the mean ± SEM, and all experiments were repeated at least three times. Student’s t-test was used to determine the statistical significance of the differences between groups. Comparative t-test was used for the clinical sample analysis. Differences with P < 0.05 were considered significant (*P < 0.05, **P < 0.01, ***P < 0.001).

To investigate the expression level of CADM2 in HCC and its clinical significance, we analyzed the high-throughput data from the GEO Database (GSE27150, GDS3091) and it confirmed that CADM2 was significantly down regulated in HCC tissues in comparison with the matched normal samples (Fig. 1a). Furthermore, survival analysis indicated a strong correlation between CADM2 mRNA expression and overall survival of patients in dataset GSM27150. In other words, the lower expression of CADM2 mRNA was associated with poor prognosis (Fig. 1b). In addition, the expression of CADM2 in HCC tissues with and without venous metastasis from GEO was also analyzed. The expression of CADM2 in liver cancer samples with venous metastasis was significantly lower than those without venous metastasis (Fig. 1c). To further confirm these results, we detected the expression of CADM2 mRNA and protein in 36 fresh HCC samples and adjacent normal liver tissues, which were collected from Heilongjiang Cancer Hospital. The results showed that CADM2 mRNA and protein expression were significantly lower in HCC tissues than those in adjacent normal liver tissues (Fig. 1d, e). These data demonstrated that CADM2 is significantly down regulated in HCC, deeming that it may serve as a tumor suppressor.

Firstly, the expression levels of CADM2 in several liver cell lines were screened. As shown in Fig. 1f, qRT-PCR analysis showed that CADM2 significantly decreased in HCC cell lines compared to normal liver cell line (HL7702). It is worth noting that the expression of CADM2 of Hep3B cells is lowest among these cell lines. To investigate the role of CADM2 in HCC progression, Wound-healing assay and Transwell assay with or without matrigel were performed in HCC cells after transfected with pEX-CADM2 or pEX-NC. The results demonstrated overexpression of CADM2 could dramatically suppress the migratory ability and invasion of HCC cells (Fig. 2a, b). It is believed that EMT process can promote the migratory ability in cancer cells. Next, we examined the EMT markers in HCC cells. Immunofluorescence analysis revealed morphological changes of HCC cells after ectopic expression of CADM2 (Additional file 4: Figure S1). Overexpression of CADM2 resulted in increased expression of E-cadherin (the epithelial cell marker) and decreased expression of Vimentin (the mesenchymal cell marker) (Fig. 2c), thus suggesting that CADM2 could regulate EMT. These results were also confirmed by Western blotting. Meantime, we observed that CADM2 also could inhibit the expression of β-catenin (Fig. 2d). β-catenin also could promote EMT progress [23]. The results demonstrate that overexpression of CADM2 significantly suppresses EMT process and inhibits migratory ability and invasion of HCC cells.

We next explored the molecular mechanisms responsible for the metastasis effect of CADM2. miRanda and TargetScan, two bioinformatics algorithms, predicted that CADM2 was a potential target gene of miR-10b. We transfected HEK293T cells with miR-10b mimics and inhibitors followed by detection of mRNA and protein expression of CADM2. We found that miR-10b overexpression led to CADM2 mRNA and protein levels decrease at 72 h post transfection (Fig. 3a, b). To determine whether CADM2 is directly targeted by miR-10b at its 3′-UTR, the luciferase reporter plasmid containing 3′-UTR fragments of CADM2 was co-transfected with miR-10b mimics and NC mimics. There are two predictive miR-10b binding sites in 3′-UTR of CADM2. We built different 3′-UTR fragments including wild type (WT) and mutant-type (MUT1, MUT2) 3′-UTR fragments (Fig. 3c). Luciferase activity assays showed that ectopic expression of miR-10b significantly decreased the luciferase activity of the WT but not that of the MUT1 or MUT2 CADM2 3′ UTR in HEK293T cells (Fig. 3d). These data indicate that miR-10b directly targets the CADM2 3′UTR, thereby reducing CADM2 expression in HCC cells.

Recent publications strongly support that miR-10b plays a vital role in the development of a variety of tumors, especially metastasis [24, 25]. Several studies find that miR-10b is able to promote invasion and migration as well as regulate EMT process in several tumor cells [26‐28]. However, there is no report on whether miR-10b can regulate EMT in HCC. To investigate whether miR-10b could regulate EMT process in HCC cells, we altered the expression of miR-10b through transient transfection of HCC cells with a miR-10b mimic or inhibitor. The expression of miR-10b in HCC cells was confirmed by qRT-PCR after transfection for 48 h (Additional file 5: Figure S2). In Immunofluorescence assay, overexpression of miR-10b inhibited E-cadherin expression and increased the Vimentin expression (Fig. 4a). To further confirm that miR-10b could promote the EMT of HCC cells, we examined the expression levels of E-cadherin and Vimentin in HCC cell lines by Western blotting (Fig. 4b).

A wound-healing assay and Transwell assay with or without matrigel showed that overexpression of miR-10b could markedly promote the migratory ability and invasion of HCC cells. Meanwhile, silencing of miR-10b suppressed the migratory ability and invasion of HCC cells (Fig. 4c, d). These results demonstrate that overexpression of miR-10b significantly promotes cell migration and invasion as well as EMT in HCC cell lines.

In order to validate the correlation between CADM2 and miR-10b expression in human HCC tissues, we firstly collected and analyzed fresh human HCC samples and corresponding non-tumor tissues from HCC resection. In the analysis of primary tumor specimens from 36 patients with HCC by qRT-PCR, miR-10b expression was found to be significantly higher in tumor tissues compared to corresponding non-tumor tissues (Fig. 5a). In addition, the level of miR-10b in liver cell lines also was detected. As we expected, the expression of miR-10b is higher in HCC cells compared to that in normal liver cells (HL7702) (Fig. 5b). Further more, higher expression of miR-10b in HCC samples was also observed in high-throughput sequencing data from TCGA compared with normal tissues (Fig. 5c). Kaplan-Meier analysis indicates that high expression of miR-10b is negatively correlated with the overall survival time of HCC patients (Fig. 5d).

We used qRT-PCR to measure CAMD2 expression in the same cohort that was used to examine the expression level of miR-10b. There was an inverse correlation between miR-10b and CADM2 expression (Fig. 5e, r² = 0.2059, P = 0.0054). Moreover, 36 HCC samples were grouped into low expression (n = 24) and high expression (n = 12) according to CADM2 expression. In the low CADM2 expression group, 21/24 of the samples showed high level of miR-10b (Fig. 5f, Additional file 6: Table S2). The results show that there is a significant negative correlation between miR-10b and CADM2 expression levels in HCC samples.

To investigate the underlying mechanisms on how miR-10b/CADM2 modulates HCC metastasis. We found out that overexpression of CADM2 suppressed the expression of FAK and p-AKT in Hep3B cells (Fig. 6a). MiR-10b overexpression increased FAK level and activated AKT pathway (Fig. 6b, c). Co-transfection miR-10b with CAMD2 partially rescued the changes of EMT markers and FAK induced by miR-10b overexpression (Fig. 6b). To further confirm miR-10b/CADM2 regulates EMT process via regulating AKT pathway, we treated HCC cells with a PI3K inhibitor, LY294002 (20 μmol/L) to block PI3K/AKT signaling. Then Western blotting and Transwell assay were performed. The results demonstrated that LY294002 partially reversed EMT process and increased migratory and invasion ability induced by miR-10b (Fig. 6c, d). These data indicated that miR-10b could activate FAK/AKT pathway to promote EMT process through inhibiting its target gene CADM2 (Fig. 6e).