Liver cancer RNA-seq data (http://kmplot.com/analysis/index.php?p=background) were extracted from the GEO, the EGA and TCGA databases. IGF2BP3 expression data associated with overall survival (OS) and relapse-free survival (RFS) were analyzed using the Kaplan-Meier plotter online tool (http://kmplot.com).

The Huh-7 and HA22T liver cancer cell lines (both adult hepatocellular carcinoma) were purchased from the American Type Culture Collection (28–30). Cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen; Thermo Fisher Scientific Inc.) supplemented with 10% fetal bovine serum (Thermo Fisher Scientific, Inc.), 1% glutamine and 1% penicillin/streptomycin (Invitrogen; Thermo Fisher Scientific, Inc.) (31,32), and cultured at 37°C in a humidified incubator [5% (v/v) CO₂]. PCR detection for mycoplasma contamination yielded negative results.

Mouse anti-ZO-1 (1:1,000; sc-33725), mouse anti-GAPDH (1:1,000; sc-47724) and mouse anti-IGF2BP3 (1:1,000; sc-365641) antibodies were purchased from Santa Cruz Biotechnology, Inc. Rabbit polyclonal argonaute-2 antibody (1:100; ab32381) was purchased from Abcam. Anti-rabbit/mouse secondary antibodies (1:5,000; A10547 and A10668, respectively) were purchased from Invitrogen; Thermo Fisher Scientific, Inc. Rabbit IgG (1:100; sc-69786) was also obtained from Santa Cruz Biotechnology, Inc. The miRNA-191-5p inhibitor was purchased from Shanghai GenePharma Co., Ltd.

The lentivirus system and standard calcium chloride transfection method were applied to generate the virus. The pWPI–IGF2BP3, pLKO.1 pLKO.1-shIGF2BP3#1 or pLKO.1-shIGF2BP3#2, pWPI-ZO-1 and pLKO.1-oemiR191-5p/pLKO.1/pLKO.1-oemiR429, or the corresponding empty control plasmids (EMD Millipore) were co-transfected into 293 cells with the pMD2G envelope plasmid and psPAX2 packaging plasmid (12259 and 12260, respectively; both Addgene, Inc.), using Lipofectamine^® 3000 (Invitrogen; Thermo Fisher Scientific, Inc.). After 8 h, the medium was replaced with fresh warm medium. The cells were then cultured in a virus room incubator for virus generation as previously decreased (33). After 48 h, the virus-containing supernatants were harvested, used immediately or stored at −80°C for later use. Huh-7 or HA22T cells were seeded into a 6-well plate (~1×10⁶/well) and infected with virus (MOI=2). Green fluorescence protein was used to monitor gene overexpression, and 1 µg/ml puromycin was to select the cells with gene knockdown or miR-191-5p overexpression. In order to downregulate miR-191-5p expression, ~1×10⁶ cells/well were seeded into a 6-well plate, and transfected with 50 pmol miR-191-5p inhibitor or NC inhibitor using Lipofectamine^® 3000, and incubated for 24 h. The expression level of miR-191-5p was monitored by reverse transcription-quantitative (RT-q) PCR to determine whether upregulation or downregulation was successful. The shRNA sequences were as follows: shIGF2BP3#1 targeting sequence, 5′-GCAGGAATTGACGCTGTATAA-3′; shIGF2BP3#2 targeting sequence, 5′-TCTGCGGCTTGTAAGTCTAT-3′; miR-191-5p inhibitor, 5′-CAGCUGCUUUUGGGAUUCCGUUG-3′; and NC inhibitor, 5′-CAGUACUUUUGUGUAGUACAA-3′.

Total RNA was extracted from Huh-7 and HA22T cells using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.). Total RNA (1 µg) was reverse transcribed into cDNA using Superscript III transcriptase (Invitrogen; Thermo Fisher Scientific, Inc.) at 42°C for 60 min and 95°C for 5 min. qPCR was conducted using a Bio-Rad CFX96 system (Applied Biosystems) with SYBR-Green to detect the mRNA expression level of a gene of interest using the following thermocycling conditions: 95°C for 3 min, followed by 35 cycles at 95°C for 15 sec, 60°C for 30 sec and 68°C for 1 min. Expression levels were normalized to the GAPDH levels using the 2^−ΔΔCq method (34). The primer sequences were as follows: miR-191-5p forward, 5′-CAACGGAATCCCAAAAGCAGCTG-3′ and reverse, 5′-TGTCGTGGAGTCGGC-3′; U6 forward, 5′-CTCGCTTCGGCAGCACA-3′ and reverse, 5′-AACGCTTCACGAATTTGCGT-3′; IGF2BP3 forward, 5′-ACTGCACGGGAAACCCATAG-3′ and reverse, 5′-ACTATCCAGCACCTCCCACT-3′; ZO-1 forward, 5′-GTGGGTAACGCCATCCTCTG-3′ and reverse, 5′-TCCGGGATTTCACCAGTGTG-3′; and GAPDH forward, 5′-TGCACCACCAACTGCTTAGC-3′ and reverse, 5′-GGCATGGACTGTGGTCATGAG-3′. All primers were purchased from Integrated DNA Technologies, Inc.

As previously described, a Transwell assay was used to determine the invasive capacity of HCC cells. Following transfection, 5×10⁴ cells/well were seeded with serum-free medium into the upper chambers of the Transwell inserts, which had been pre-coated with diluted Matrigel (Corning Inc.) at 37°C for 4 h. Subsequently, 750 µl media supplemented with 10% FBS was pipetted into the lower chambers, and the cells were incubated at 37°C [5% (v/v) CO₂] for 18 h. After being fixed with methanol at room temperature for 20 min, the invasive cells were stained with 0.1% (w/v) crystal violet at room temperature for 15 min. The average number of cells from three randomly selected microscopic areas using an inverted light microscope (magnification, ×100; Olympus Corporation) were counted and for quantification, and the number of invasive cells was calculated using ImageJ software 1.8 (National Institutes of Health).

Cells were lysed to extract total proteins using RIPA buffer (Wuhan Boster Biological Technology, Ltd.,). The protein concentration was measured using a BCA Protein assay kit (ab102536; Abcam) and ultraviolet spectrophotometry. Proteins (30 µg/well) were separated using 6–10% SDS/PAGE gels and then transferred onto PVDF membranes (EMD Millipore). After blocking with 5% milk at room temperature for 1.5 h, the membranes were incubated with primary antibodies at 4°C overnight, and subsequently with HRP-conjugated secondary antibodies (detail in the Reagents and materials part) at room temperature for 2 h. An enhanced chemiluminescent substrate (35055; Thermo Fisher Scientific, Inc.) was then used to detect and visualize the signals. The relative expression levels were quantified using the Image Lab soft 4.1 (Bio-Rad Laboratories, Inc.) with GAPDH as the internal reference.

Following transfection, cells were lysed in ice-cold lysis buffer supplemented with RNase inhibitor. Following centrifugation, 500 µl of the supernatant was clarified using protein A/G beads for 1 h and incubated with an argonaute-2 antibody at 4°C overnight. RNA was extracted using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol and subjected to RT-qPCR analysis.

A 500 bp fragment of the ZO-1 3′ untranslated region (UTR), with wild-type or mutant IGF2BP3 or miRNA-responsive elements, was cloned into the psiCHECK-2 vector (Promega Corporation) downstream of the Renilla luciferase open reading frame. HA22T and Huh-7 cells were seeded into 24-well plates, and the cDNA was transfected with Lipofectamine^® 3000 transfection reagent according to the manufacturer's instructions. PRL-TK was used as an internal control that served as the baseline control response. Luciferase activity was measured at 48 h post-transfection using a dual-luciferase reporter assay (Promega Corporation) according to the manufacturer's protocol. The relative luciferase activity was normalized to Renilla luciferase activity.

Statistical analysis was performed using SPSS software, version 23.0 (IBM Inc.). The Mann-Whitney U test was conducted to compare two groups of continuous data. The Kruskal-Wallis test with the Dunn-Bonferroni post hoc test was applied to multiple groups of continuous data. All cellular experiments were conducted in three technical replicates. Quantitation Results are presented as the mean ± SD. Statistical significance for cell experiments was determined using the independent-sample t-test. P<0.05 was considered to indicate a statistically significant difference.

The datasets, which included gene expression associated with OS and RFS, were retrieved from the GEO, EGA and TCGA, and were analyzed online (http://kmplot.com). The results revealed that patients with a high IGF2BP3 expression level (50% of the top IGF2BP3 expression samples) exhibited lower OS and RFS rates than those with low IGF2BP3 expression (50% of the lower IGF2BP3 expression samples) (Fig. 1A and B), suggesting that IGF2BP3 may be involved in HCC progression. Furthermore, IGF2BP3 expression was determined in patients at different grades or stages of HCC using data from TCGA database. The results demonstrated that patients with higher grades or stages of the disease possessed higher IGF2BP3 expression levels (Fig. 1C and D). Taken together, these data are consistent with those previously published, which suggest that IGF2BP3 promotes HCC progression (13).

To further demonstrate the function of IGF2BP3 in HCC progression, its effects on cell invasiveness were assessed by IGF2BP3 shRNA knockdown or overexpression with IGF2BP3-cDNA in HA22T and Huh-7 cells, respectively. As shown in Fig. 2A and B, IGF2BP3 knockdown with IGF2BP3-shRNA#1 markedly decreased the invasive ability of HA22T cells. To avoid the off-target effect of IGF2BP3-shRNA, a second shIGF2BP3#2 plasmid was constructed to repeat the experiment, and the results were consistent with those obtained with IGF2BP3-shRNA#1 (Fig. S1A and B). By contrast, the overexpression of IGF2BP3 markedly increased the invasive capacity of Huh-7 cells (Fig. S1C and D).

To elucidate the mechanisms responsible for the effects of IGF2BP3 on HCC cell invasiveness, the expression of ZO-1 was examined, a protein which is associated with cellular invasiveness in a variety of cancers, such as breast, pancreatic and lung cancer (27,35,36). IGF2BP3 knockdown in HA22T cells resulted in increased expression of ZO-1 (Fig. 2C). Conversely, the overexpression of IGF2BP3 in Huh-7 cells decreased the expression of ZO-1 (Fig. 2D). To verify whether increased levels of ZO-1 could reduce cellular invasion capacity, a lentivirus system was used to overexpress ZO-1 and western blot analysis was used confirm expression in Huh-7 cells (Fig. S1E). The results of Transwell assay revealed that ZO-1 overexpression markedly decreased Huh-7 cell invasiveness (Fig. 2E). These results suggest that IGF2BP3 affects HCC cell invasiveness by altering ZO-1 expression.

A reverse assay was then employed to examine the roles of ZO-1 in the IGF2BP3-induced increase in HCC cell invasiveness. The results revealed that the ZO-1 overexpression partially blocked the IGF2BP3-induced increase in HCC cell invasiveness (Fig. 2E-F). Collectively, the results shown in Fig. 2A-F demonstrate that IGF2BP3 may increase HCC cell invasiveness by suppressing ZO-1 expression.

The present study then aimed to reveal the mechanisms responsible for the suppressive effects of IGF2BP3 on ZO-1 expression. It has been previously reported that IGF2BP3 acts as an RNA-binding protein by recognizing specific motifs to determine targeted RNA fate (37), and that IGF2BP3 can modulate invasion-associated gene transcripts by promoting Ago2-mRNA interactions (23). Therefore, the potential regulation of ZO-1 by Ago2-mRNA interactions was investigated by detecting ZO-1 mRNA in the Ago2 complex using an RIP assay. ZO-1 mRNA co-precipitation with Ago2 was decreased in the IGF2BP3-depleted cells (Fig. 3A). Subsequently, to determine which miRNAs mediated ZO-1 expression through the IGF2BP3-Ago2 complex, online miRNA software (http://www.targetscan.org) was used to screen six potential candidates and detect miRNA expression from the Ago2-antibody pull-down complex between the negative control (NC) cells and IGF2BP3-cDNA cells. The results revealed that the miR-191-5p and miR429 levels were significantly higher in the IGF2BP3-overexpressing cells than in the NC cells in the Ago2 complex (Fig. 3B). The roles of the two miRNA candidates in ZO-1 reduction were then investigated; an increase in miR191-5p alone decreased ZO-1 expression (Fig. S2A and B). However, overexpressing miR429 in HA22T cells (Fig. S2C) had no impact on ZO-1 protein expression (Fig. S2D). It was thus hypothesized that the miR191-5p regulatory effects were associated with the 3′UTR overlap of the IGF2BP3 binding sites, and that IGF2BP3 synergized miR191-5p via the RISC. To further investigate this hypothesis, a rescue experiment was performed to determine the role of miR191-5p in the IGF2BP3-induced increase in HCC cell invasiveness. The findings revealed that miR191-5p knockdown with an miR191-5p inhibitor (Fig. S2E) partially reversed the IGF2BP3-induced increase in HCC cell invasion and decreased ZO-1 expression (Fig. 3C-D). Taken together, the data shown in Fig. 3A-D suggest that IGF2BP3 decreases ZO-1 expression by cooperating with the miR191-5p-Ago2 complex via the RISC.

To further determine the mechanisms of the regulatory effects of the miR191-5p-IGF2BP3-Ago2 complex on ZO-1 mRNA degradation. The 3′UTRs of ZO-1 mRNA were first analyzed with a consensus sequence of ‘AAUGU’ and IGF2BP3-RNA was found to crosslink the overlapping miR191-5p target sites in the of ZO-1 mRNA 3′UTRs (23). A luciferase assay was then applied to determine whether IGF2BP3-miR191-5p directly targeted the overlap of ZO-1 mRNA 3′UTR to downregulate its expression. The wild-type and mutant miRNA-target sites were first cloned into the psicheck2 vector to construct the reporter plasmids (Fig. 4A). As expected, increasing miR191-5p resulted in decreased luciferase activity in HA22T cells transfected with the wild-type, but not the mutant ZO-1 3′UTR (Fig. 4B). Subsequently, other reporter plasmids were constructed with psicheck2 vectors carrying the wild-type and mutant IGF2BP3-target sites (Fig. 4C). The luciferase assay revealed that increasing IGF2BP3 expression resulted in decreased luciferase activity in Huh-7 cells transfected with the wild-type ZO-1 3′UTR, but not in the mutant 3′UTR. In addition, attenuating IGF2BP3 significantly increased the luciferase activity in HA22T cells transfected with the wild-type, but not in the mutant ZO-1 3′UTR (Fig. 4D). Taken together, the results presented in Fig. 4A-D suggest that IGF2BP3 can promote the function of miR191-5p by binding the overlap target sites and enhancing Ago2-mRNA interactions.