Human colorectal cancer cell lines (SW480 and HT29) and 293T cells were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). The cells were cultured in L-15 or DMEM medium (Hyclone; Cytiva) containing 10% fetal bovine serum (FBS) (Gibco; Thermo Fisher Scientific, Inc.). On reaching 80% confluency, SW480 and HT29 cells in the logarithmic growth phase were seeded into low-adhesion 12-well plates (1×10⁵ cells/well) and maintained in stem cell culture medium: 25 µl B27 (1:50; Gibco; Thermo Fisher Scientific, Inc.), 20 µl EGF (20 ng/ml; Invitrogen; Thermo Fisher Scientific, Inc.), 15 µl basic fibroblast growth factor (bFGF; 10 ng/ml; Invitrogen; Thermo Fisher Scientific, Inc.) and 940 µl L-15 medium. The stem cell culture medium was replenished every 72 h, and the cells were cultured continuously for 14 days; the resulting SW480-3D and HT29-3D cell microspheres were maintained in stem cell culture medium and harvested for subsequent experimentation as required.

A total of 118 primary CRC and paired-adjacent tissue samples were collected from patients with CRC at the Affiliated Hospital of Zunyi Medical University (Zunyi, China), who underwent radical resection in a blinded manner between January 2015 and December 2015. The patient data are presented in Table I. CRC diagnosis was based on histopathological features. No patient underwent radiotherapy or chemotherapy preoperatively. Following surgery, the tissue samples were immediately stored at −80°C for reverse transcription-quantitative (RT-q)PCR analysis. A 5-year follow-up survey of the patients was performed to determine their survival status. The present study was reviewed and approved by the Ethics Review Committee of the Affiliated Hospital of Zunyi Medical University (approval no. [2015] 1-040), and written informed consent was obtained from all patients.

Total RNA from tissues and cells was extracted using TRIzol^® reagent according to the manufacturer's instructions (Takara Biotechnology Co., Ltd.), and reverse transcribed into cDNA using the Prime Script RT kit (Takara Biotechnology Co., Ltd.). Subsequent qPCR detection was performed using the 7500 Real-Time PCR System (Applied Biosystems; Thermo Fisher Scientific, Inc.) with a 20 µl reaction mixture comprising 10 µl qPCR SYBR Green Mix (Takara Biotechnology Co., Ltd.), 0.8 µl each of the forward and reverse primers, 2 µl cDNA and diethyl pyrocarbonate-treated H₂O up to the final volume. The qPCR conditions were: Initial denaturation at 95°C for 30 sec, followed by 40 cycles of denaturation at 95°C for 5 sec, annealing at 60°C for 30 sec and melting curve analysis. Each test set included three duplicate wells, and the experiment was repeated three times. Relative gene expression was normalized to that of GAPDH or U6, and calculated using the 2^−ΔΔCq method (42). The primers were designed and synthesized by General Biosystems, Inc., the sequence of which are as follows: miR-8063 forward, 5′-TGCGGTCAAAATCAGGAGTCGGGG-3′ and reverse, 5′-CCAGTGCAGGGTCCGAGGT-3′; U6 forward, 5′-GCTCGCTTCGGCAGCACA-3′ and reverse, 5′-AACGCTTCACGAATTTGCGTG-3′; hnRNPAB forward, 5′-AAGAAGTCTATCAGCAGCAGCAGTATG-3′ and reverse, 5′-CTCCACCTCCACCACCACCTC-3′; and GAPDH forward, 5′-ATGACATCAAGAAGGTGGTGAAGCAGG-3′ and Reverse, 5′-GCGTCAAAGGTGGAGGAGTGGG-3′.

Total cellular protein was extracted, and its concentration determined, using the Protein Extraction kit and the BCA protein concentration kit (both Beyotime Institute of Biotechnology), respectively. Protein samples (40 µg) were separated using 10% SDS-PAGE and transferred to a PVDF membrane. After blocking with 5% skim milk for 1 h at room temperature, the membrane was incubated overnight at 4°C with anti-hnRNPAB (cat. no. 14813-1-AP; 1:2,000), anti-Wnt3a (cat. no. 26744-1-AP; 1:2,000), anti-Wnt5a (cat. no. 55184-1-AP; 1:2,000), anti-β-catenin (cat. no. 17565-1-AP; 1:2,000) and anti-β-tubulin (cat. no. 10094-1-AP; 1:5,000) (all ProteinTech Group, Inc.). The following day, the membrane was washed three times with PBS containing 0.1% Tween-20, and incubated with a horseradish peroxidase-conjugated antibody solution (cat. no. SA00001-2; 1:10,000; ProteinTech Group, Inc.) for 1 h at room temperature. Protein bands were visualized using Super ECL plus super-sensitive luminescent solution (Bio-Rad Laboratories, Inc.) and exposed using X-ray film. Quantity One software v4.6.6 (Bio-Rad Laboratories, Inc.) was used to quantify band intensities.

The target gene of hnRNPAB (miR-8063) was predicted using TargetScan (http://www.targetscan.org/vert_71/) bioinformatics software. The wild-type plasmid pmirGLO/hnRNPAB-3′UTR and the mutant plasmid pmirGLO/hnRNPAB-3′UTR-mut were constructed by General Biosystems, Inc. 293T cells (1×10⁵/well) were seeded into 24-well plates 24 h before transfection. Cells were co-transfected with pmirGLO/hnRNPAB-3′UTR or pmirGLO/hnRNPAB-3′UTR-mut and miR-8063 mimics using Lipofectamine^® 3000 (Invitrogen; Thermo Fisher Scientific, Inc.). After transfection for 48 h, the dual-luciferase reporter assay kit (Beyotime Institute of Biotechnology) was used according to the manufacturer's instructions, and the relative light unit (RLU) value was determined using the FLx800 fluorescence analyzer (BioTek Instruments, Inc.). The firefly RLU value was normalized to that of Renilla luciferase.

A soft agar colony-formation assay was used to evaluate the CSC characteristics of HT29-3D microspheres, and the self-renewal ability of CRCSCs (SW480CSCs and HT29CSCs) after miR-8063 overexpression. First, two different concentrations of soft agarose (0.6 and 1.2%) were prepared. The 3D microspheres and corresponding parent cells were digested into a single cell suspension using Accutase enzyme (PAN-Biotech) and 0.25% trypsin (Hyclone; Cytiva), respectively. The digested cells were then resuspended in a medium containing 10% FBS (1×10³ cells/ml), and mixed with 1.2% agarose at a 1:1 ratio. Then, 3 ml of the mixture was added to each 6-cm-diameter glass dishes, and left to solidify to form the lower agar layer. Medium and 0.7% agarose were then mixed at a 1:1 ratio, and 0.5 ml cell suspension was added to the mix, which was then added atop the previously prepared 1.2% agarose to form the upper agar layer. Finally, the culture dishes were incubated at 37°C for 2 weeks. The number of colonies in each plate was counted using a light microscope.

For parent cells (SW480 and HT29), a plate colony-formation assay was used to detect CSC characteristics after hnRNPAB overexpression. Cells were seeded into 6-well plates (1×10³/well) in 1 ml medium containing 10% FBS, and cultured at 37°C for 2 weeks. Next, 1 ml 4% paraformaldehyde was added as a fixative at room temperature for 15 min, and 1 ml Giemsa dye was then added for 15 min at room temperature. The number of colonies was manually counted using a light microscope (>50 cells are considered a clone). Colony-formation rate=number of clones/number of seeded cells ×100%.

A total of 12 male specific pathogen-free-grade NOD/SCID mice (weight, 20–25 g; age, 6 weeks) and 44 male BALB/C nude mice (weight, 15–20 g; age, 6 weeks) were used in the experiment. All mice were housed at 25°C, 50% humidity and in specific-pathogen-free conditions with a 12/12-h light/dark cycle. Sterile food and water were provided daily. NOD/SCID mice and nude mice were purchased from the Animal Experimental Center, Institute of Radiology Medicine, Chinese Academy of Medical Sciences (Tianjin, China). NOD/SCID mice were used to evaluate the tumorigenicity of HT29-3D microspheres in vivo. HT29-3D microspheres and HT29 cells were harvested and counted. Then, 1×10⁷ cells were collected and resuspended in 1 ml medium; ~100 µl suspension (containing 1×10⁶ cells) was subcutaneously injected into the left armpits of NOD/SCID mice (six mice in each group). Similarly, 5×10⁶ SW480 and HT29 cells (overexpressing hnRNPAB), SW480CSCs and HT29CSCs cells (overexpressing the miR-8063), and the corresponding NC cells, were subcutaneously injected into the left armpits of the nude mice. When the tumors had reached a maximum diameter of 15 mm, the mice were anesthetized with an intraperitoneal injection of 1% Pentobarbital (50 mg/kg), and then sacrificed by cervical dislocation. Tumor volume was calculated using the following formula: Tumor volume = ½ (length × width²) (43).

The expression of CSC markers was detected using flow cytometry. Cells in the logarithmic growth phase were homogenized into a single cell suspension. Then, 1×10⁶ cells were resuspended in 100 ml PBS containing 5% bovine serum albumin and 10 µl fluorophore-conjugated primary anti-CD44-PE (cat. no. PE-6506), anti-CD133-PE (cat. no. PE-62403) and the corresponding negative control antibodies (cat. no. PE-48642) (all ProteinTech Group, Inc.). The cells were mixed and incubated for 10 min in the dark at 4°C. After centrifugation (500 × g, 5 min at room temperature), the cells were washed three times with PBS and resuspended in 200 µl PBS each. Finally, NovoExpress software 1.2.4 (Novocyte; ACEA Biosciences, Inc.) was used to detect the percentage of fluorescence-positive cells.

The lentiviruses overexpressing hnRNPAB (hnRNPAB-GFP-PURO), miR-8063-mimic-GFP-PURO and miR-8063-inhibitor-GFP-PURO were purchased from Hanbio Biotechnology Co., Ltd. The miR-8063-inhibitor-GFP-PURO lentiviral vector sequence was as follows: 5′-UUCGGGGCUGAGACUAAAACU-3′. Cells (1×10⁵) were seeded into 12-well plates 24 h before infection. The virus was used to infect the cells according to the optimal MOI value obtained in the pre-experiment: Lentivirus hnRNPAB-GFP-PURO (MOI of 30 for both SW480 and HT29), miR-8063-mimic-GFP-PURO (MOI of 20 for SW480CSCs, and 30 for HT29CSCs), and miR-8063-inhibitor GFP-PURO (MOI 20 for both SW480 and HT29). Then, 8 mg/ml Polybrene (Hanheng Biological Technology Co., Ltd.) and enhanced infection solution (Shanghai GeneChem Co., Ltd.) were added at 37°C for 12 h, and then replaced with fresh medium. After a further 72 h, the cells were collected, and the effect of gene overexpression or silencing was determined by RT-qPCR or western blotting.

The data are expressed as the mean ± standard deviation. Multigroup comparisons were conducted by one-way ANOVA followed by Tukey's post hoc test. The χ² test was used to determine the association between miR-8063 expression and patient clinical characteristics. The Kaplan-Meier method was used to assess the association between overall survival rate and the expression level of miR-8063. P<0.05 was considered to indicate a statistically significant difference.

SW480-3D and HT29-3D microspheres were generated using a suspension culture of human CRC cells (SW480 and HT29; Fig. 1). A colony formation assay was performed to verify the self-renewal properties of the HT29-3D microspheres, and the results indicated that the colony-formation rates of HT29-3D microspheres and HT29 cells (45.67±9%) were significantly higher than those of the parent cells (16.98±5%) (Fig. 2A and B) (P<0.01). Next, the tumorigenicity of HT29-3D microspheres was detected using an in vivo tumorigenicity assay in NOD/SCID mice. The tumor volume of the mice inoculated with HT29-3D microspheres was significantly larger than that of the HT29 cell group, suggesting that HT29-3D microspheres had greater tumorigenicity than the parent cells (Fig. 2C and D) (P<0.01). Finally, the results of flow cytometry showed that the positive expression rates of CD44 in HT29-3D microspheres and HT29 cells were 37.14±1.62 and 9.37±2.28%, respectively. The corresponding positive expression rates of CD133 were 38.40±1.74 and 3.64±0.95%, respectively. These findings demonstrate significantly higher expression of CD44 and CD133 in HT29-3D microspheres compared with the parent cells (Fig. 2E and F) (P<0.01). The results suggest that HT29 microspheres exhibit CSC characteristics. SW480-3D microspheres with CSC characteristics were identified and named SW480CSCs in our previous study (11). Similarly, HT29-3D microspheres were termed HT29CSCs.

RT-qPCR detection revealed that the relative expression levels of miR-8063 in SW480 and SW480CSCs were 1.03±0.04 and 0.45±0.06, respectively. miR-8063 expression in HT29 and HT29CSCs were 0.99±0.11 and 0.32±0.12, respectively (Fig. 3A) (P<0.01). Furthermore, relative hnRNPAB mRNA expression in SW480CSCs and HT29CSCs, compared with the parental cells, was 2.6- and 3.2-fold, respectively, as detected by RT-qPCR. (Fig. 3B). Moreover, the expression of hnRNPAB protein in SW480, SW480CSCs, HT29 and HT29CSCs were 0.74±0.04, 1.04±0.06, 0.68±0.03 and 0.98±0.07, respectively (Fig. 3C and D). These results indicate that compared with the parent cells, the expression level of miR-8063 in SW480CSCs and HT29CSCs was significantly decreased, while hnRNPAB expression was significantly increased (P<0.01).

The patient population comprised 65 men and 53 women (mean age, 58 years; age range, 30–82 years). Compared with the adjacent tissues, miR-8063 expression in CRC tissues was significantly downregulated (Fig. 4A; P<0.001). Next, the expression levels were classified as low and high according to the median value (0.62), and association between miR-8063 expression and patient clinicopathological characteristics was determined. As shown in Table I, low miR-8063 expression was significantly associated with advanced TNM stage (P=0.004), tumor infiltration (P=0.008), vascular invasion (P=0.010) and lymph node metastasis (P=0.020). No association was found between miR-8063 expression and age, sex, differentiation and tumor site. These data indicated that low expression of miR-8063 was closely associated with the malignant behaviors of CRC. The relationship between miR-8063 expression and the overall survival rate of patients was estimated using the Kaplan-Meier method; the overall survival rate of CRC patients with high miR-8063 expression was lower than those with low miR-8063 expression, suggesting an association between high expression and poorer prognosis (Fig. 4B; P=0.023).

To verify whether there is a direct relationship between miR-8063 and hnRNPAB, bioinformatics software (TargetScan Human) was used to predict common binding sites between the two molecules. The results showed that the hnRNPAB gene possesses a potential miR-8063 binding sequence in its 3′UTR region (Fig. 5A). Furthermore, the dual-luciferase reporter assay revealed that compared with the other groups, the relative RLU value of the pmirGLO/hnRNPAB-3′UTR+miR-8063 mimics group was significantly lower (P<0.01), confirming that hnRNPAB is the direct target gene of miR-8063 (Fig. 5B and C) (P<0.01).

To investigate the effect of hnRNPAB on the stemness of CRC cells, a lentiviral vector overexpressing hnRNPAB (hnRNPAB-GFP-PURO) and the corresponding NC (NC-GFP-PURO) were transfected into SW480 and HT29 cells, and the experimental groups (SW480-hnRNPAB and HT29-hnRNPAB) and the NC group (SW480-NC and HT29-NC) were established. First, overexpression efficiency was tested. The hnRNPAB mRNA expression in the SW480-hnRNPAB and HT29-hnRNPAB groups was 3.5-fold and 3.3-fold, respectively, compared with that of the NC groups (Fig. 6A). The expression of hnRNPAB protein also showed a similar trend (Fig. 6B and C) (P<0.01). These results indicated that the hnRNPAB overexpression model was successfully constructed.

Furthermore, the colony formation rates of the SW480-hnRNPAB, SW480-NC, HT29-hnRNPAB and HT29-NC groups were 18.8±0.42, 5.6±0.21, 16.1±0.53 and 5.7±0.31%, respectively, indicating a significant increase in the colony-formation ability of SW480 and HT29 cells following hnRNPAB overexpression (Fig. 6D and E) (P<0.01).

The results of tumor formation in nude mice showed that all groups formed tumors, though the SW480-hnRNPAB and HT29-hnRNPAB groups exhibited increased tumor volumes compared with the NC groups (Fig. 6F and G). Similarly, the SW480-hnRNPAB and HT29-hnRNPAB groups possessed a significantly greater tumor volume than the NC-treated mice (Fig. 6H-J). These findings suggested that the tumorigenicity of CRC cells was significantly increased following hnRNPAB overexpression.

In addition, the proportions of CD44⁺ cells in the SW480-NC, SW480-hnRNPAB, HT29-NC and HT29-hnRNPAB groups were determined by flow cytometry, and were 27.05±2.57, 64.68±4.40, 5.08±2.72 and 33.27±5.27%, respectively. The corresponding proportions of CD133⁺ cells were 19.19±5.49, 87.88±6.18, 6.37±2.26 and 27.62±5.89% respectively (Fig. 6K and L) (P<0.01). These results suggested that the positive expression rates of CD44 and CD133 in the SW480-hnRNPAB and HT29-hnRNPAB groups were significantly greater than those of the corresponding NC groups.

The relative protein expression levels of Wnt3a in the SW480-NC, SW480-hnRNPAB, HT29-NC and HT29-hnRNPAB groups were 0.27±0.03, 0.87±0.04, 0.52±0.04 and 0.91±0.03, respectively; and those of Wnt5a were 0.37±0.04, 0.92±0.05, 0.28±0.05 and 1.01±0.06, respectively. The corresponding protein expression levels of β-catenin in the four groups were 0.61±0.06, 0.96±0.06, 0.59±0.04 and 0.99±0.15, respectively. Thus, compared with the NC groups, the expression of Wnt/β-catenin pathway proteins in the experimental groups was significantly increased (Fig. 7A and B) (P<0.01).

SW480CSCs and HT29CSCs (obtained from SW480 and HT29 cells by suspension culture) were transfected with a lentiviral vector overexpressing miR-8063 and the corresponding NC. The experimental groups (SW480CSCs-miR-8063 and HT29CSCs-miR-8063) and NC groups (SW480CSCs-NC and HT29CSCs-NC) were established. Compared with the NC group, the SW480CSCs-miR-8063 and HT29CSCs-miR-8063 groups exhibited 2.7-fold and 3.3-fold relative expression of miR-8063, respectively, as detected by RT-qPCR (Fig. 8A; P<0.01).

The colony formation rates of the SW480CSCs-miR-8063, SW480CSCs-NC, HT29CSCs-miR-8063 and HT29CSCS-NC groups were 17.43±4.30, 42.59±3.48, 39.52±7.28, and 74.93±8.52%, respectively. Compared with those in the NC groups, the colony formation rates in the experimental groups were significantly decreased (Fig. 8B and C) (P<0.01). These results showed that following miR-8063 overexpression, the colony-formation ability of SW480CSCs and HT29CSCs was significantly decreased.

Following subcutaneous inoculation, the subcutaneous tumor growth of nude mice was observed weekly, and a growth curve was plotted (Fig. 8D and E). While all the groups formed tumors, the tumor volumes in the SW480CSCs-miR-8063 and HT29CSCs-miR-8063 groups were significantly lower than those of the corresponding NC-treated mice (Fig. 8F and G), suggesting that the tumorigenic ability of SW480CSCs and HT29CSCs in nude mice was significantly reduced by the overexpression of miR-8063.

The positive expression rates of CD44 in the SW480CSCs-miR-8063, SW480CSCs-NC, HT29CSCs-miR-8063 and HT29CSCs-NC groups were 3.35±1.87, 24.86±3.27, 6.42±1.3 and 35.06±2.40%, respectively. The corresponding positive expression rates of CD133 were 31.33±2.61, 87.91±4.99, 10.47.±2.52 and 33.07±3.86%, respectively (Fig. 8H and I) (P<0.01). These results indicate that the expression of CD44⁺ and CD133⁺ in the SW480CSCS-miR-8063 and the HT29CSCS-miR-8063 groups was significantly lower than that in the NC groups.

The mRNA expression levels of hnRNPAB in the SW480CSCS-NC, SW480CSCS-miR-8063, HT29CSCS-NC and HT29CSCS-miR-8063 groups were 1.03±0.11, 0.68±0.04, 0.96±0.05 and 0.53±0.02, respectively (Fig. 9A; P<0.01). The same trend was observed for protein expression of hnRNPAB (Fig. 9B and C; P<0.01). These results indicated that overexpression of miR-8063 inhibited the hnRNPAB expression.

Next, the relative protein expression of Wnt3a in the SW480CSCS-NC, SW480CSCS-miR-8063, HT29CSCS-NC and HT29CSCS-miR-8063 groups was 0.68±0.09, 0.47±0.04, 1.10±0.13 and 0.65±0.047, respectively. Concerning Wnt5a, the relative protein expression in the four groups was 0.49±0.03, 0.29±0.06, 0.85±0.15 and 0.52±0.06, respectively. The corresponding relative expression of the β-catenin protein was 0.68±0.04, 0.46±0.08, 0.71±0.05 and 0.58±0.03, respectively. Compared with the corresponding NC group, the expression levels of the three proteins in the SW480CSCs-miR-8063 and HT29CSCs-miR-8063 groups were significantly decreased (Fig. 9B and C; P<0.01). These results suggest that the overexpression of the miR-8063 gene inhibited the Wnt/β-catenin signaling pathway.

miR-8063 was silenced in SW480 and HT29 cells using the lentiviral-mediated RNA interference (RNAi) technique, and experimental groups (SW480-sh-miR-8063 and HT29-sh-miR-8063) and NC groups (SW480-sh-NC and HT29-sh-NC) transfected with NC virus were established. Compared with the NC groups, the relative expression of miR-8063 in the experimental groups was significantly decreased (Fig. 10A; P<0.01). The results of RT-qPCR showed that compared with the NC group, the relative expression of hnRNPAB mRNA in the SW480-sh-miR-8063 and HT29-sh-miR-8063 groups was 2.9-fold and 5.7-fold, respectively (Fig. 10B; P<0.01). Similarly, the expression of hnRNPAB protein (detected by western blotting) was also significantly upregulated compared with that of the NC groups (Fig. 10C and D). These results indicated that hnRNPAB expression was upregulated following miR-8063 silencing. Next, we examined whether the upregulation of hnRNPAB expression induced by silencing miR-8063 was accompanied by the activation of the Wnt/β-catenin signaling pathway. Compared with the NC groups, the relative expression levels of Wnt3a, Wnt5a and β-catenin in the SW480CSCs-miR-8063 and HT29CSCs-miR-8063 groups were significantly increased (Fig. 10C and D) (P<0.01). These findings suggest that miR-8063 silencing promoted the expression of hnRNPAB, resulting in the activation of the Wnt/β-catenin signaling pathway.