All patients with GC in this study were treatment-naïve prior to surgery. Primary GC tissues and corresponding normal gastric tissues were obtained from 20 patients undergoing surgical resection at the Department of Surgery of the First Affiliated Hospital of Nanchang University (Nanchang, China) from December, 2016 to November, 2017. Following resection, the fresh tissue samples were frozen in liquid nitrogen and stored at −80°C. The clinicopathological characteristics of the patients were confirmed by two experienced pathologists and are summarized in Table I. This study was approved by the Independent Ethics Committee of the First Affiliated Hospital of Nanchang University and complied with the Declaration of Helsinki (Approval no. #021-2017). All patients agreed to participate in this study and provided written informed consent.

Three GC cell lines (AGS, MGC803 and BGC823), the normal gastric epithelial cell line, GES-1, and 293 cells were obtained from Sun Yat-Sen University Cancer Center (Guangzhou, China), and the GC cell line, SGC7901, was purchased from the Shanghai Institute of Cell Biology, China Academy of Sciences (Shanghai, China). The cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS) (both from HyClone, Logan, UT, USA) in a humidified chamber at 37°C in an atmosphere containing 5% CO₂.

RNA was isolated from fresh human tissues or harvested cells using TRIzol reagent (Invitrogen/Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions, as previously described (24). Complementary DNA was synthesised using a PrimeScript RT Reagent kit (Takara, Ohtsu, Japan), and quantitative PCR (qPCR) was performed using the ABI 7500 Real-Time Fast PCR System (Applied Biosystems, Foster City, CA, USA) with SYBR-Green qPCR SuperMix (Invitrogen/Thermo Fisher Scientific). Primers targeting miR-381, miR-489 and CUL4B were purchased from GenePharm (Shanghai, China). U6 or GAPDH were used as internal controls, and all results were calculated using the 2^−ΔΔCq method (25). The primer sequences are presented in Table II.

Briefly, the cells were lysed in lysis buffer consisting 50 mM Tris/HCl (pH 7.5), 150 mM NaCl, 1 mM dithiothreitol (DTT), 1 mM EDTA, 0.5% Nonidet P-40 (NP-40), 0.2 mM phenylmethylsulfonyl fluoride (PMSF), 10 µM pepstatin A and 1 mM leupeptin. Equal amounts of clear cell lysates were separated by 8 or 10% SDS-PAGE polyacrylamide gel electrophoresis on 8 or 10% gels and electroblotted onto polyvinylidene difluoride membranes (Bio-Rad Laboratories, Hercules, CA, USA). The membranes were blocked with a 5% milk solution at room temperature, before being incubated at 4°C with the following primary antibodies overnight: CUL4B (diluted 1:1,000; 60151-1-Ig; Proteintech, Rosemont, IL, USA), β-catenin (1:1,000; #9562), cyclin D1 (1:1,000; #2978), c-Myc (1:1,000; #5605) (all from Cell Signaling Technology, Danvers, MA, USA), E-cadherin (1:1,000; ab1416), N-cadherin (1:1,000; ab18203), Vimentin (1:1,500; ab8978), β-actin (1:2,000; ab119716) (all from Abcam, Cambridge, MA, USA) and GAPDH (1:2,000; 14C10; Santa Cruz Biotechnology, Dallas, TX, USA). The following day, the membranes were incubated with corresponding horseradish peroxidase-conjugated secondary antibodies (anti-rabbit or anti-mouse; Cell Signaling Technology; #7074, 1:3,000 and #7076, 1:4,000) for 1 h at room temperature. The signal was detected with super sensitive regent (Thermo Fisher Scientific), and β-actin or GAPDH was used as an internal control. Quantification of the protein bands were analysed using ImageJ software (Rawak Software, Stuttgart, Germany).

miR-381 and miR-489 mimics and the corresponding negative control, CUL4B small interfering RNA (siRNA), control siRNA, lentiviral vectors of hsa-miR-381, hsa-miR-489 and hsa-miR-negative control were purchased from GenePharm. The siRNA sequences were as follows: siRNA-NC, 5′-UUCUCC GAACGUGUCACGUTT-3′; and siRNA-CUL4B, 5′-CCACCC AGAAGUCAUUAAUTT-3′. The transfection of oligonucleotides into target cells was performed using TurboFect reagent (Thermo Fisher Scientific) according to the manufacturer’s instructions. Certain cells, as indicated, were transduced using lentivirus-containing supernatant for 24 h in the presence of 2.5 µg/ml polybrene. Following infection, puromycin selection was used to establish stable cell lines. An expression vector carrying a CUL4B sequence lacking the 3′-UTR (CUL4B-no UTR) was constructed by inserting the coding sequence into the psiCHECK-2 vector (Promega, Madison, WI, USA). RT-qPCR and western blot analysis were performed to confirm successful transfection. The data shown were derived from triplicate samples.

The identification of upstream regulatory miRNAs targeting CUL4B was performed using two independent miRNA databases: miRanda (http://www.microrna.org/) and TargetScan (http://www.targetscan.org). The full-length 3′-UTR of the CUL4B gene and a variant sequence were amplified by PCR and cloned into the pGL3 luciferase reporter vector (Promega). The wild-type plasmid (CUL4B-Wt-3′-UTR) contained the full-length CUL4B 3′-UTR, including the sequences complementary to miR-381 and miR-489. The mutant plasmids (CUL4B-Mut-3′-UTR) carried variant 3′-UTR sequences created by replacing ‘CUUGUAU’ with ‘GAACAUA’ in the sequence complementary to miR-381 or ‘UAUGAUGU’ with ‘ACACUACA’ in that complementary to miR-489. The luciferase reporter assay was performed as previously described. 293 cells were seeded at 2×10⁵ cells per well in 12-well plates and transiently co-transfected with 1 µg reporter plasmid and 100 nM miR-381 mimic, miR-489 mimic, or negative control using TurboFect reagent for 48 h. Luciferase activity was measured using a Dual-Luciferase Reporter Assay kit (E1910; Promega) according to the manufacturer’s instructions. Renilla luciferase activity was used for normalization, and all experiments included 3 biological replicates.

To assess long-term cell viability, a CCK-8 assay (Dojindo Molecular Technologies, Rockville, MD, USA) was used according to the manufacturer’s instructions as previously described (26). Following transfection for 24 h, the cells were seeded in 96-well plates. Cell viability was measured by the addition of WST-8 at a final concentration of 10% and measuring the absorbance of samples at 450 nm in a microplate reader (SpectraMax M5e; Molecular Devices, Sunnyvale, CA, USA) every 24 h for 5 days. The data was derived from triplicate samples and are presetned as means ± standard errors of the mean (SEM).

The colony formation assay was carried out as previously described (24). The MGC803 and BGC823 cells (500 per well) were seeded in a 60-mm dish 24 h post-transfection and cultured in RPMI-1640 medium containing 10% FBS for 10 days. Colonies comprising >50 cells were then fixed with 10% formaldehyde for 30 min and stained with 8.0% crystal violet (#R40052; Thermo Fisher Scientific) for a further 30 min at room temperature. Each experiment was repeated 3 times.

To evaluate cell motility, the indicated cells, at a density of 4.5×10⁵ per well, were seeded in 6-well plates and serum-starved for 24 h once fully confluent. A sterile 200-µl pipette tip was then used to scratch the culture to form a wound. The cells were subsequently carefully washed with phosphate-buffered saline and cultured in serum-free medium. Images were captured at 0 and 48 h after scratching to evaluate wound closure under a microscope (#CKX41; Olympus America, Inc., Center Valley, MA, USA). Percentage wound closure was calculated by comparing the wound area at a given time to that at 0 h. These experiments were performed at least 3 times.

The Transwell assay was performed as described previously (26). For the migration assay, the indicated cells were seeded in the upper chambers of Transwell inserts with non-coated membranes (8-µm pore size; BD Biosciences, Franklin Lakes, NJ, USA). For the invasion assay, cells were suspended in 200 µl serum-free medium and seeded in the upper chambers of Transwell inserts pre-coated with Matrigel. Medium (500 µl) containing 15% FBS was added to the lower chambers as a chemoattractant. Following 72 h of incubation at 37°C, cells that had invaded the bottom surface of the membranes were fixed and stained with 10% crystal violet at room temperature. These experiments were performed using 3 biological replicates.

A total of 24 (6 weeks old; weighing, 18-21 g) female NOD SCID mice were purchased from Jackson Laboratories (Bar Harbor, ME, USA). Mice were supplied with free sterilized water and food at 22±2°C with 40-70% humidity. The mice were randomly allocated to 1 of 4 groups (6 mice in each) and inoculated with 5×10⁶ MGC803 cells stably expressing miR-negative control, hsa-miR-381, or hsa-miR-489 by subcutaneous injection into both anterior flanks. Tumour growth in two dimensions was monitored every 3 days using electronic digital callipers (Thermo Fisher Scientific). Tumour volume was calculated with the following formula: Tumour volume (mm³) = (length × width²)/2. The mice were euthanised, and tumours were harvested and weighed. All animal experiments were approved by the Animal Ethics Committee of Nanchang University. The maximum tumour diameter allowed per tumour was 1.5 cm.

Statistical analysis was performed using SPSS 17.0 software (SPSS Inc., Chicago, IL, USA). All data are presented as the means ± SEM, and each experiment was repeated at least 3 times. Comparisons between 2 groups were made using a Student’s t-test, while one-way ANOVA with a post hoc Dunnett or Bonferroni correction test was used when analysing >2 groups. The association between miRNA expression and CUL4B protein expression was analysed by Pearson’s correlation coefficient. A P-value <0.05 was considered to indicate a statistically significant difference.

Our preliminary immunohistochemical data revealed that CUL4B was upregulated in GC tissues compared with paired adjacent normal gastric tissues (data not shown). In the present study, we found that 15 of the 20 GC samples (75%) examined exhibited a higher CUL4B protein expression than the matched adjacent non-cancerous gastric tissues (Fig. 1A; P<0.01). To investigate the underlying mechanisms of CUL4B upregulation, we searched for potential miRNAs that target CUL4B using bioinformatics analysis (http://www.microrna.org/microrna/home.do). We screened out several miRNAs that may regulate CUL4B expression according to the SVR score and conserved status. Moreover, with TargetScan (www.Targetscan.org) for the second screening, miR-381 and miR-489 attracted our attention. The reasons for the selection of miR-381 and miR-489 for further investigation in this study were as follows: i) Two algorithms both predicted that miR-381 and miR-489 were potential regulatory miRNAs that could target CUL4B; and ii) these two miRNAs have been reported to be involved in cancer progression (27-29). As shown in Fig. 1B, potential miR-381 and miR-489 target sites are present in the CUL4B 3′-UTR. As the expression patterns of miRNAs are often the opposite of those of their target genes, we examined miR-381 and miR-489 expression in the above-mentioned 20 tissue sample pairs by RT-qPCR. As clearly illustrated in Fig. 1C, the miR-381 and miR-489 levels were consistently downregulated in the GC tissues compared with the non-cancerous controls. Correlation analysis indicated that the expression of miR-381 and miR-489 was negatively correlated with CUL4B protein expression in the 20 clinical GC samples (Fig. 1D; r=−0.571 and r=−0.518, respectively; P<0.01). The expression of these miRNAs and CUL4B was also examined in the AGS, BGC823, MGC803 and SGC7901 human GC cells and in GES-1 normal gastric epithelial cells. Of note, CUL4B protein expression was much higher in the AGS, BGC823 and MGC803 cells, but lower in the SGC7901 cells, and was inversely associated with miR-381 and miR-489 expression (Fig. 1E and F). Collectively, these data suggest that the downregulation of miR-381 and miR-489 may play an important role in the pathogenesis of GC by affecting CUL4B expression.

To determine whether CUL4B is a direct target of miR-381 and miR-489 in GC, fragments of its 3′-UTR containing the predicted binding sites of these miRNAs or variants of these sequences were separately subcloned into the pGL3 vector (Fig. 2A). Luciferase reporter assays were then performed to verify the binding of the CUL4B 3′-UTR by these two miRNAs. As depicted in Fig. 2B, the miR-381 and miR-489 levels were significantly increased in the MGC803 and BGC823 cells upon transfection. We also found that the ectopic expression of miR-381 or miR-489 exerted a significant inhibitory effect on the luciferase activity of the CUL4B-Wt-3′-UTR plasmid in the 293 cells, but had no marked effect on that of the CUL4B-Mut-3′-UTR (P<0.01; Fig. 2C and D). As shown in Fig. 2E and F, miR-381 or miR-489 upregulation decreased CUL4B protein, but not mRNA expression in the MGC803 and BGC823 cells, suggesting that these miRNAs regulate CUL4B levels principally through translational repression. Collectively, these results strongly support the hypothesis that CUL4B is a direct target of miR-381/miR-489, and that they specifically and directly suppress CUL4B expression by binding to its 3′-UTR.

To determine the functional effects of CUL4B on GC cells, we examined cellular phenotypes following its knockdown using siRNA. Successful knockdown was confirmed by RT-qPCR (Fig. 3A) and western blot analysis (Fig. 3H). As clearly shown in Fig. 3B and C, the silencing of CUL4B suppressed GC cell growth. Consistent with this finding, a significant decrease in the colony formation ability was detected in the cells in which CUL4B was knocked down when compared with the negative control (P<0.01; Fig. 3D and E). Moreover, CUL4B silencing markedly diminished the migratory and invasive abilities of the MGC803 and BGC823 cells (P<0.01; Fig. 3F and G). In previous studies, CUL4B was found to positively regulate the Wnt/β-catenin pathway in pancreatic cancer and hepatocellular carcinoma, and the abnormal activation of this pathway has been shown to be involved in tumourigenesis (13,30). Therefore, in this study, we examined whether Wnt/β-catenin signalling was affected by the silencing of CUL4B. As shown in Fig. 3H, the expression of β-catenin, c-Myc and cyclin D1 was downregulated in the MGC803 and BGC823 cells transfected with CUL4B siRNA. These data suggest that the silencing of CUL4B exerts a marked tumour suppressive effect on GC tumourigenesis via the inactivation of the Wnt/β-catenin pathway.

Given that miR-381 and miR-489 are frequently downregulated in GC (22,23), we expressed these miRNAs ectopically to explore their biological functions. As illustrated in Fig. 4A and B, miR-381 and miR-489 over-expression reduced cell viability. Consistent with this, we observed a significant reduction in the number of colonies formed by the miRNA transfectants compared with the negative control (Fig. 4C and D). Subsequently, to determine whether these miRNAs can inhibit tumour growth in vivo, the MGC803 cells were infected with lentivirus expressing miR-381, miR-489, or negative control miRNA, and puro-mycin selection was used to establish lines stably expressing these constructs. These cells (5×10⁶ per mouse) were then injected into the anterior flanks of nude mice. Compared with the miR-381 group, the miR-NC group developed significantly larger tumours (P<0.01; Fig. 4E). Similarly, miR-489 overexpression resulted in the formation of markedly smaller tumour nodules and attenuated the growth of tumour xenografts in the nude mice (P<0.01; Fig. 4F). Taken together, these results clearly suggest that the ectopic expression of miR-381/miR-489 strongly delays GC cell growth in vitro and in vivo.

We then evaluated the effects of miR-381 and miR-489 on GC cell migration and invasion using wound healing and Transwell invasion assays. The results of the wound healing assay revealed that the cells overexpressing miR-381/miR-489 took a much longer time to close the scratch wound than the controls (P<0.05; Fig. 5A and B). Moreover, miR-381 or miR-489 overexpression significantly suppressed the invasive abilities of the MGC803 and BGC823 cells (P<0.01; Fig. 5C and D). EMT is a critical process contributing to the initiation of cancer cell invasion; therefore, we also examined EMT markers in these two cell lines following miR-381 or miR-489 upregulation. Surprisingly, the levels of E-cadherin were increased and those of N-cadherin and Vimentin were reduced due to the elevated expression of miR-381 or miR-489 (Fig. 5E). Furthermore, the combination of miR-381 and miR-489 overexpression suppressed cell growth, migration and invasion more markedly (data not shown). The simultaneous overexpression of miR-381 and miR-489 exerted a more potent tumour-suppressive effect on the GC cells (data not shown). These data indicate a critical role of miR-381/miR-489 in cell migration, invasion and EMT, apart from cell proliferation.

To further confirm that the functional effects of miR-381 and miR-489 in GC are mediated by the inhibition of CUL4B, we overexpressed a CUL4B sequence lacking the 3′-UTR in the miRNA-transfected cells. As shown in Fig. 6A, the restoration of CUL4B expression partially negated the effects of miR-381 and miR-489 on the protein levels of CUL4B and components of the Wnt/β-catenin pathway. Furthermore, CCK-8, colony formation and Transwell invasion assays demonstrated that the restoration of CUL4B expression also counteracted the inhibitory effects of miR-381 and miR-489 on GC cell proliferation and invasion (Fig. 6B-F). Taken together, these results further demonstrate that the altered CUL4B expression is a main mediating factor in the functions of both miR-381 and miR-489.