HCC pathologically confirmed tumor tissues and adjacent healthy tissues (>5 cm from the tumor margin) were obtained from 38 patients with HCC (31-85 years old) who had undergone surgical treatment at People's Hospital of Dongying between January 2011 and June 2014. The present study was approved by People's Hospital of Dongying Ethics Committee (Dongying, China). Written informed consent was obtained from all patients.

Human normal liver cell line (THLE-2) and HCC cell lines (SK-HEP-1 and Hep3B) were purchased from the American Type Culture Collection. HCC cell lines (Huh-7 and Huh-1) were purchased from the Japanese Collection of Research Bioresources Cell Bank. All cells were maintained at 37˚C with 5% CO₂ in DMEM (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% FBS (Invitrogen; Thermo Fisher Scientific, Inc.).

DUXAP8 small interfering (si)RNA (si-DUXAP8#1, 5'-AGTTCAGTGTATTTGATAATA-3'; si-DUXAP8#2, 5'-GATTTGGTTTCAGAATCAAAG-3'; si-DUXAP8#3, 5'-GATGGTGTTGTACCACCTATA-3') and scramble siRNA control [si-negative control (NC): 5'-TTCTCCGAACGTGTCACGTTT-3'] were purchased from Shanghai GenePharma Co., Ltd. miR-9-3p mimic (5'-ATAAAGCTAGATAACCGAAAGT-3'), scramble mimic control (miR-NC; 5'-TTCTCCGAACGTGTCACGTTT-3'), miR-9-3p inhibitor (5'-ACTTTCGGTTATCTAGCTTTAT-3') and scramble inhibitor control (anti-miR-NC; 5'-CAGTACTTTTGTGTAGTACAA-3') were purchased from Applied Biosystems; Thermo Fisher Scientific, Inc. Furthermore, to overexpress DUXAP8 or IGF1R, the full length sequence of DUXAP8 or IGF1R was inserted into the pcDNA3.1 vector (Invitrogen; Thermo Fisher Scientific, Inc.) to generate pcDNA3.1-DUXAP8 (DUXAP8) and pcDNA3.1-IGF1R (IGF1R), respectively. pcDNA3.1 empty vector was used as negative control. SK-HEP-1 and Huh-7 cells in 6-well plates (2x10⁵ cells/well) were transfected with oligonucleotides (50 nM) and plasmids (2 µg) using Lipofectamine^® 2000 (Invitrogen; Thermo Fisher Scientific, Inc.). Following incubation for 48 h, the transfected cells were harvested and utilized for subsequent experiments.

Total RNA was extracted from clinical tissues and HCC cells using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.). Subsequently, total RNA was reverse transcribed into cDNA using M-MLV reverse transcriptase (Thermo Fisher Scientific, Inc.) including 5x buffer, 10 mM dNTPs, and 50 µmol oligo dT primers (cat. no. 18418012; Thermo Fisher Scientific, Inc.). The RT reaction was performed at 70˚C for 10 min followed by 42˚C for 1 h, according to the manufacturer's protocol. DUXAP8 and IGF1R mRNA expression levels were measured via qPCR using SYBR Premix Ex Taq (Takara Biotechnology Co., Ltd.). The thermocycling conditions were as follows: Denaturation at 95˚C for 10 min, followed by 40 cycles of denaturation at 95˚C for 30 sec, annealing at 60˚C for 30 sec and extension at 72˚C for 1 min. Has-miR-9-3p-specific TaqMan primer (5'-ATAAAGCTAGATAACCGAAAGT-3') was purchased from Applied Biosystems (cat. no. 4427975; Thermo Fisher Scientific, Inc.; Assay ID, 002231) for synthesis of cDNA. miR-9-3p expression levels were measured via qPCR using a TaqMan MicroRNA Assay Kit (Applied Biosystems; Thermo Fisher Scientific, Inc.). The thermocycling conditions were as follows: 16˚C for 30 min, 42˚C for 30 min, 85˚C for 5 min and 4˚C hold. The following primers were used for qPCR: DUXAP8 forward, 5'-AGACGCCATGGAACAT-3' and reverse, 5'-AAGCGGAGACCTGAGGAG-3'; GAPDH forward, 5'-AGAAGGCTGGGGCTCATTTG-3' and reverse, 5'-AGGGGCCATCCACAGTCTTC-3'; IGF1R forward, 5'-TTTCCCACAGCAGTCCACCTC-3' and reverse, 5'-AGCATCCTAGCCTTCTCACCC-3'; miR-9-3p forward, 5'-TCTTTGGTTATCTAGCTGTAT-3' and reverse, 5'-GAACATGTCTGCGTATCTC-3'; and U6 forward, 5'-GCTTCGGCAGCACATATACTAAAA-3' and reverse, 5'-CGCTTCACGAATTTGCGTGTCAT-3'. Moreover, miRNA and mRNA expression levels were quantified using the 2^-ΔΔCq method (31) and normalized to the internal reference genes U6 and GAPDH, respectively.

Cell proliferation was detected by performing a Cell Counting Kit-8 (CCK-8) assay (Beyotime Institute of Biotechnology) according to the manufacturer's protocol. Briefly, transfected SK-HEP-1 and Huh-7 cells (2.5-5x10³ cells/well) were inoculated into 96-well plates and incubated for 48 h at 37˚C. Subsequently, CCK-8 (10 µl) solution was added to each well and cultured at 37˚C for 0, 24, 48 or 72 h. The optical density value of each well was measured at a wavelength of 450 nm using a Varioskan Flash Microplate Reader (Thermo Fisher Scientific, Inc.).

Cell migration and invasion were detected using 24-well Transwell chambers (BD Biosciences) according to the manufacturer's protocol. Transfected SK-HEP-1 and Huh-7 cells were resuspended in serum-free DMEM overnight. Subsequently, cells (5x10⁴) were plated into the upper chamber, which was pre-coated with Matrigel^® at 4˚C overnight and subsequently maintained at 37˚C for 4-5 h (BD Biosciences) for the cell invasion assay. DMEM supplemented with 10% FBS (Invitrogen; Thermo Fisher Scientific, Inc.) was plated in the lower chamber. Following incubation for 24 h at 37˚C, cells in the upper chamber were removed using cotton swabs. Migratory or invading cells were fixed using methanol for 30 min at 4˚C and stained with 0.1% crystal violet solution for 20 min at 37˚C, followed by observation using an inverted fluorescence microscope (magnification, x100; Nikon Corporation).

Western blotting was performed to measure the protein expression levels of IGF1R and EMT-related proteins (E-cadherin, N-cadherin and vimentin) in SK-HEP-1 and Huh-7 cells. Total protein was isolated from cells using pre-cold RIPA buffer (Beyotime Institute of Biotechnology) including protease inhibitor. Total protein was quantified using a BCA protein assay kit (Pierce; Thermo Fisher Scientific, Inc.). Proteins (50 µg) were separated via 10% SDS-PAGE and electrotransferred to nitrocellulose membranes (EMD Millipore). The membranes were blocked with 5% non-fat milk at room temperature for 1 h. Subsequently, the membranes were incubated overnight at 4˚C with primary antibodies targeted against: IGF1R (1:1,000; cat. no. 3027), Snail (1:1,000; cat. no. 3879), Slug (1:1,000; cat. no. 9585), E-cadherin (1:1,000; cat. no. 14472), N-cadherin (1:500; cat. no. 14215), vimentin (1:1,000; cat. no. 5741) and GAPDH (1:1,000; cat. no. 5174; all from Cell Signaling Technology, Inc.). The membranes were incubated with a horseradish peroxidase-labeled anti-rabbit IgG secondary antibody (1:1,000; sc-2027; Santa Cruz Biotechnology, Inc.) at room temperature for 1 h. Protein bands were visualized using an ECL detection kit (Pierce; Thermo Fisher Scientific, Inc.) and analyzed with Quantity One v4.6.2 software (Bio-Rad Laboratories, Inc.). GAPDH was used as the loading control.

The target gene of DUXAP8 or miR-9-3p was predicted using StarBase v2.0 (http://starbase.sysu.edu.cn/starbase2/) and TargetScan (www.targetscan.org) online prediction software, respectively. Based on the bioinformatics prediction, a partial DUXAP8 fragment containing the wild-type (WT) or mutant (MUT) miR-9-3p binding site, or the IGF1R 3'untranslated region (UTR) sequence containing the WT or MUT miR-9-3p binding site were amplified from HCC cells via PCR with 1 unit AmpliTaq DNA polymerase (Thermo Fisher Scientific, Inc.) using the following thermo cycling conditions: Pre-denaturation at 94˚C for 10 min; 30 cycles at 94˚C for 30 sec, 53-57˚C for 30 sec and 72˚C for 45 sec; and final extension at 72˚C for 10 min). The PCR product was subsequently cloned into the psiCHECK-2 vector (Promega Corporation) using the restriction endonuclease cleavage sites of XhoI and NotI (Promega Corporation). The following primers were used: DUXAP8-WT forward, 5'-CCGCTCGAGAACACTAATTGTAGACTATG-3' and reverse, 5'-ATAAGAATGCGGCCGCTATTGATGAGGATTTTCAAT-3'; DUXAP8-MUT forward, 5'-TCAAAACCCAGAAAACCATACAAGGACGATTTGTGA-3' and reverse, 5'-AACCCAGAAAACCATACAAGGACGATTTGTGAATT-3'; IGF1R-WT forward, 5'-CCGCTCGAGTAGTCAGTTGACGAAGATCT-3' and reverse, 5'-ATAAGAATGCGGCCGCTAGCTACACTCCAAAGGGAA-3'; and IGF1R-MUT forward, 5'-TTAGGACACCTGTTTACTAGAGGAACCGCAAATATGC-3' and reverse, 5'-GACACCTGTTTACTAGAGGAACCGCAAATATGCCAA-3'. SK-HEP-1 and Huh-7 cells in 24-well culture plates (2x10⁵ cells/well) were co-transfected using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc.) with 100 ng of DUXAP8-WT, DUXAP8-MUT, IGF1R-WT or IGF1R-MUT and 100 nM miR-9-3p mimic or miR-NC. Following incubation for 48 h, luciferase activities were assessed using a Dual-Luciferase reporter assay kit (cat. no. E1910; Promega Corporation), according to the manufacturer's protocol. Firefly luciferase activity was normalized to that of Renilla luciferase.

The StarBase Pan-Cancer platform (http://starbase.sysu.edu.cn/panCancer.php) was used to analyze the correlation between DUXAP8, miR-9-3p and IGF1R in liver hepatocellular carcinoma (LIHC) (32,33).

To investigate the relationship between DUXAP8 and miR-9-3p, RIP was performed using the Magna RIP RNA-Binding Protein Immunoprecipitation kit (EMD Millipore) according to the manufacturer's protocol. Briefly, at 80% confluence, SK-HEP-1 and Huh-7 cells were harvested and lysed in RIP lysis buffer (EMD Millipore). Subsequently, 100 µl cell extract was incubated with RIP buffer including magnetic beads conjugated with anti-Argonaute2 (Ago2) antibody (1:1,000; cat. no. 03-248) or normal mouse IgG antibody (1:1,000, cat. no. 12-371; both from EMD Millipore) overnight at 4˚C. Proteinase K (Invitrogen; Thermo Fisher Scientific, Inc.) was applied to digest the samples for 30 min at 55˚C, and the isolated RNAs were utilized for RT-qPCR analysis of DUXAP8 and miR-9-3p expression levels.

All statistical analyses were performed using SPSS software (version 18; SPSS, Inc.). Data are presented as the mean ± SD. The χ² test was used to analyze the association between DUXAP8 expression levels and clinicopathological features in HCC. Overall survival rates were evaluated using the Kaplan-Meier method with the long rank test applied for comparisons. The relationship among DUXAP8, miR-9-3p and IGF1R was analyzed using Pearson's correlation coefficient. Comparisons between two groups were analyzed using a paired Student's t-test. Comparisons among multiple groups were analyzed using one-way ANOVA followed by Tukey's post hoc test. P<0.05 was considered to indicate a statistically significant difference. Experiments were repeated at least three times.

To investigate the function of DUXAP8 in HCC, its expression was detected via RT-qPCR. Compared with adjacent healthy tissues, the expression of DUXAP8 was significantly increased in 38 HCC tumor tissues (Fig. 1A). DUXAP8 expression levels were also significantly elevated in HCC cell lines (SK-HEP-1, Huh-7, Hep38 and Huh-1) compared with the human liver cell line (THLE-2; Fig. 1B), with the highest expression levels observed in SK-HEP-1 and Huh-7 cells. Therefore, SK-HEP-1 and Huh-7 cells were selected for subsequent experiments. The association between DUXAP8 expression and clinicopathological factors in the 38 patients with HCC are presented in Table I. DUXAP8 expression was significantly associated with tumor size, TNM stage and metastasis (P<0.05), but was not significantly associated with age, sex, HbsAg and cirrhosis. Moreover, the overall survival curves suggested that DUXAP8 expression was inversely associated with the prognosis of patients with HCC (Fig. 1C). The results indicated that DUXAP8 may serve as a novel biomarker for the prognosis of HCC.

To assess the function of DUXAP8 in HCC cells, DUXAP8 knockdown was performed. The expression of DUXAP8 was significantly decreased in si-DUXAP8-transfected SK-HEP-1 and Huh-7 cells compared with si-NC-transfected cells (Fig. 2A). Since si-DUXAP8#1 exhibited the highest knockdown inefficiency, it was selected for subsequent experiments. Therefore, the role of DUXAP8 in proliferation, migration, invasion and EMT was investigated by knocking down DUXAP8 expression in HCC cells. The CCK-8 and Transwell assay results suggested that DUXAP8 knockdown significantly inhibited SK-HEP-1 and Huh-7 cell proliferation (Fig. 2B and C), migration (Fig. 2D) and invasion (Fig. 2E) compared with si-NC. Subsequently, the effect of DUXAP8 on HCC cell EMT was investigated. The western blotting results indicated that DUXAP8 knockdown significantly increased E-cadherin protein expression levels, but significantly decreased Snail, Slug, N-cadherin and vimentin protein expression levels compared with si-NC (Fig. 2F), which suggested that DUXAP8 promoted SK-HEP-1 and Huh-7 cell EMT. Collectively, the results suggested that DUXAP8 enhanced HCC cell proliferation, migration, invasion and EMT.

Increasing evidence has indicated that lncRNAs could exert their role by interacting with miRNAs (34-36). Therefore, bioinformatics-based target prediction analysis using StarBase was performed to identify the target miRNAs of DUXAP8. miR-9-3p was found to bind with DUXAP8 (Fig. 3D). To further assess the interaction between DUXAP8 and miR-9-3p, SK-HEP-1 and Huh-7 cells were co-transfected with DUXAP8-WT or DUXAP8-MUT reporter vector and miR-NC or miR-9-3p mimics, and the subsequent luciferase activities were measured. The results suggested that miR-9-3p overexpression significantly decreased the luciferase activity of the DUXAP8-WT reporter compared with miR-NC, but miR-9-3p overexpression had no significant effect on the luciferase activity of the DUXAP8-MUT reporter compared with miR-NC (Fig. 3E and F). Moreover, in accordance with the bioinformatics analysis and luciferase assay, the RIP assay results suggested that DUXAP8 and miR-9-3p were significantly enriched in Ago2 pellets of SK-HEP-1 and Huh-7 cell extracts compared with the IgG control group (Fig. 3G and H). Meanwhile, the transfection efficiencies of miR-9-3p mimics and DUXAP8 overexpression were detected (Fig. S1). Moreover, miR-9-3p expression was significantly decreased in HCC tumor tissues and cell lines compared with adjacent healthy tissues and a human liver cell line, respectively (Fig. 3A and B). miR-9-3p expression levels were inversely correlated with DUXAP8 expression levels in HCC tumor tissues (Fig. 3C). Subsequently, miR-9-3p expression levels were detected in DUXAP8-transfected and si-DUXAP8-transfected SK-HEP-1 and Huh-7 cells. The RT-qPCR results suggested that DUXAP8 overexpression significantly downregulated miR-9-3p expression compared with pcDNA3.1, whereas DUXAP8 knockdown significantly upregulated miR-9-3p expression levels compared with si-NC (Fig. 3I and J), suggesting that manipulation of DUXAP8 expression altered the expression of miR-9-3p. The results suggested that DUXAP8 interacted with miR-9-3p to repress its expression.

To further investigate the mechanism underlying DUXAP8 in HCC progression, rescue experiments were performed by transfecting SK-HEP-1 and Huh-7 cells with si-NC, si-DUXAP8, si-DUXAP8 + anti-miR-NC or si-DUXAP8 + miR-9-3p inhibitor. The RT-qPCR results indicated that DUXAP8 knockdown significantly increased miR-9-3p expression levels compared with si-NC, which were significantly reversed by co-transfection with miR-9-3p inhibitor in SK-HEP-1 and Huh-7 cells (Fig. 4A). The transfection efficiency of miR-9-3p inhibitor was examined (Fig. S1). Furthermore, compared with si-NC, DUXAP8 knockdown significantly inhibited SK-HEP-1 and Huh-7 cell proliferation, migration and invasion, and miR-9-3p knockdown reversed DUXAP8 knockdown-mediated effects (Figs. 4B-E, S2A and S2B). The western blotting results suggested that miR-9-3p inhibition significantly decreased si-DUXAP8-induced increases in E-cadherin protein expression levels, and decreases in Snail, Slug, N-cadherin and vimentin protein expression levels in SK-HEP-1 and Huh-7 cells (Fig. 4F-I). Collectively, the results suggested that DUXAP8 may exert its carcinogenic effects partially by regulating miR-9-3p.

Previous reports have demonstrated that miRNAs may exert their function by interacting with mRNAs (37-39). Therefore, TargetScan was used to analyze the potential downstream target of miR-9-3p. IGF1R was predicted to possess binding sites with miR-9-3p (Fig. 5A). To validate the bioinformatics prediction, SK-HEP-1 and Huh-7 cells were co-transfected with IGF1R-WT or IGF1R-MUT reporter plasmids and miR-NC or miR-9-3p mimics. The luciferase assay suggested that compared with miR-NC, miR-9-3p overexpression significantly inhibited the luciferase activity of the IGF1R-WT reporter, but had no significant effect on the luciferase activity of the IGF1R-MUT reporter (Fig. 5B and C). Subsequently, IGF1R expression in miR-9-3p mimics- or miR-9-3p inhibitor-transfected SK-HEP-1 and Huh-7 cells was detected. The RT-qPCR and western blotting results suggested that miR-9-3p expression altered the expression of IGF1R, as evidenced by miR-9-3p overexpression significantly decreasing IGF1R expression levels compared with miR-NC, and miR-9-3p knockdown significantly increasing IGF1R expression levels compared with anti-miR-NC in SK-HEP-1 and Huh-7 cells (Fig. 5D-G). Moreover, IGF1R expression was significantly upregulated and negatively correlated with miR-9-3p expression in HCC tumor tissues (Fig. 5H and I). Collectively, the results suggested that IGF1R was a target of miR-9-3p in HCC cells.

To further investigate whether miR-9-3p could exert its function via targeting in HCC cells, SK-HEP-1 and Huh-7 cells were transfected with miR-NC, miR-9-3p mimics, miR-9-3p mimics + pcDNA3.1 or miR-9-3p mimics + pcDNA3.1-IGF1R. The RT-qPCR and western blotting results indicated that miR-9-3p overexpression significantly decreased the expression of IGF1R in SK-HEP-1 and Huh-7 cells compared with miR-NC, which was reversed by co-transfection with pcDNA3.1-IGF1R (Fig. 6A-C). In addition, according to the analysis performed using the Pan-Cancer platform of StarBase, the expression correlation of DUXAP8, IGF1R and miR-9-3p in LIHC samples was assessed. The expression of miR-9-3p was positively correlated with that of DUXAP8 or IGF1R, and DUXAP8 expression was positively correlated with that of IGF1R (Fig. S3). Simultaneously, the transfection efficiency of IGF1R overexpression was measured (Fig. S1). Compared with miR-NC, miR-9-3p overexpression significantly inhibited SK-HEP-1 and Huh-7 cell proliferation, migration and invasion, which was reversed by IGF1R overexpression (Figs. 6D-G, S2C and S2D). In addition, compared with miR-NC, miR-9-3p overexpression significantly increased E-cadherin protein expression levels, and decreased Snail, Slug, N-cadherin and vimentin protein expression levels, which were reversed by co-transfection with pcDNA3.1-IGF1R (Fig. 6H and I), indicating that IGF1R abolished miR-9-3p-mediated inhibition of SK-HEP-1 and Huh-7 cell EMT. The results indicated that miR-9-3p repressed HCC development by modulating IGF1R.

According to the aforementioned results, it was hypothesized that DUXAP8 could exert its carcinogenic effects via the DUXAP8/miR-9-3p/IGF1R regulatory signaling pathway. Furthermore, DUXAP8 expression was positively correlated with IGF1R expression in HCC tumor tissues (Fig. 7A). Subsequently, whether DUXAP8 could regulate the expression of IGF1R via miR-9-3p was investigated. The RT-qPCR and western blotting results suggested that DUXAP8 knockdown reduced the expression of IGF1R compared with si-NC, and co-transfection with miR-9-3p inhibitor reversed si-DUXAP8-mediated inhibitory effects on IGF1R expression in SK-HEP-1 and Huh-7 cells (Fig. 7B-D). Collectively, the results suggested that DUXAP8 may serve as a molecular sponge to sequester miR-9-3p from IGF1R in HCC cells.