A total of 70 patients (age, 42–77 years; 24 males and 46 females) with CRC undergoing surgical resection between March 2017 and May 2018 in The Third Affiliated Hospital of Southern Medical University were enrolled in the current study. None of the patients underwent preoperative chemoradiotherapy. After all patients signed informed consent forms, the tumor tissues and healthy tissues adjacent to the tumor were collected and stored at −80°C for subsequent experiments. The Ethics Committee of The Third Affiliated Hospital of Southern Medical University approved the study.

CRC cell lines, HT29 (cells were confirmed using short tandem repeat profiling), HCT116, HCT8, LOVO, SW480 and SW620, and wild-type colon epithelial cell line, FHC, were purchased from the American Type Culture Collection. HT29, HCT116, HCT8, LOVO, SW480 and SW620 cells were maintained in RPMI medium (Gibco; Thermo Fisher Scientific, Inc.) containing 10% FBS (Gibco; Thermo Fisher Scientific, Inc.) in a 37°C incubator (Thermo Fisher Scientific, Inc.) with 5% CO₂. FHC cells were cultured in DMEM/F12 medium (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% FBS, 10 mM HEPES, 5 µg/ml insulin, 10 ng/ml cholera toxin, 5 µg/ml transferrin and 100 ng/ml hydrocortisone (all Gibco; Thermo Fisher Scientific, Inc.) in a 37°C incubator with 5% CO₂.

Total RNA was extracted from CRC tissues and cells using TRIzol^® reagent (Sigma-Aldrich; Merck KGaA) and transcribed into the corresponding cDNA at 37°C for 15 min followed by 10 sec at 85°C using PrimeScript RT reagent kit with gDNA Eraser (Takara Biotechnology Co., Ltd.) and miR-X miRNA First-Strand Synthesis kit (Takara Biotechnology Co., Ltd.), according to the manufacturer's instructions. The qPCR reaction was conducted using SYBR Green PCR Master mix (Takara Biotechnology Co., Ltd.) and MiR-X miRNA RT-qPCR TB Green kit (Takara Biotechnology Co., Ltd.), according to the manufacturer's instructions. The PCR conditions were as follows: Initial denaturation at 95°C for 5 min, followed by 40 cycles at 95°C for 10 sec, 60°C for 30 sec and 72°C for 10 sec, and a final extension at 72°C for 10 min. The sequences of primers were: circ-FOXO3 forward (F), 5′-GTGGGGAACTTCACTGGTGCTAAG-3′ and reverse (R), 5′-GTCGTATCCAGTGCAGGGT-3′ (21); miR-543 F, 5′-CTCCCTCCCGAATTTGAAG-3′ and R, 5′-GTCAGAGGGAGAGGTCAG-3′ (22); large tumor suppressor kinase 1 (LATS1) F, 5′-CCACCCTACCCAAAACATCTG-3′ and R, 5′-CGCTGCTGATGAGATTTGAGTAC-3′ (23); GAPDH F, 5′-CATGAGAAGTATGACAACAGCCT-3′ and R, 5′-AGTCCTTCCACGATACCAAAGT-3′ (21); and U6 F, 5′-GCGCGTCGTGAAGCGTTC-3′ and R, 5′-GTGCAGGGTCCGAGGT-3′ (22). GAPDH and U6 were used as internal references. The relative RNA expression was analyzed using the 2^−ΔΔCq method (24).

Total RNA was extracted using TRIzol^® reagent (Sigma-Aldrich; Merck KGaA), and then 5 µg RNA was incubated with or without 3 U/µg RNase R (Epicentre Biotechnologies; Illumina, Inc.) at 37°C for 15 min. The resulting RNA was purified using an RNeasy MinElute cleaning kit (Qiagen China Co., Ltd.) according to the manufacturer's instructions. Then, the expression levels of circ-FOXO3 and FOXO3 were determined using RT-qPCR assay. The primers of FOXO3 used in this study were: F, 5′-CGGCTAGCTGCGCCTTGGCTTTATAACT-3′ and R, 5′-GGCTCGAGCCCTCCTTCACTGCTACTGG-3′ (25).

HT29 and HCT116 cells (1×10⁷) were lysed using cell fractionation buffer (Thermo Fisher Scientific, Inc.), followed by centrifuged at 16,000 × g for 5 min at 4°C to separate nuclear and cytoplasmic fractions. The obtained nuclear and cytoplasm were used for extracting RNA using a PARIS kit (Thermo Fisher Scientific, Inc.), following the manufacturer's protocol. The extracted RNA was subjected to RT-qPCR to determine the expression of circ-FOXO3.

The pcDNA circ-FOXO3, miR-543 mimic (5′-AAACAUUCGCGGUGCACUUCUU-3′), miR-543 inhibitor (5′-UACUUAAUGAGAAGUUGCCCGUGUUUUUUUCGCUUUAUUUGUGACGAAACAUUCGCGGUGCACUUCUUUUUCAGUAU-3′), small interfering (si)-LATS1 (5′-GCAATCAGTTAACCGCAAA-3′) and indicated controls, including pcDNA vector, miRNA mimic negative control (mimic NC) (5′-UUUGUACUACACAAAAGUACUG-3′), miRNA inhibitor NC (inhibitor NC) (5′-ACUACUGAGUGACAGUAGA-3′), siRNA NC (si-NC) (5′-GCACAGTTAACCGCATAAA-3′) were obtained from Guangzhou RiboBio Co., Ltd. For cell transfection, cells (1×10⁶) were inoculated into 6-well plates and maintained for 24 h at 37°C, followed by transfection with 100 nM pcDNA circ-FOXO3, miR-543 mimic, miR-543 inhibitor or si-LATS1 using Lipofectamine^® 2000 (Invitrogen; Thermo Fisher Scientific, Inc.). After 48 h of transfection, HT29 and HCT116 cells were harvested for subsequent experiments. The transfection efficiency was determined using RT-qPCR and cell fluorescence.

For cell fluorescence, cells were fixed with 4% paraformaldehyde for 10 min at room temperature and incubated with Cy3-labeled miR-543 probe in hybridization buffer at 37°C overnight. The nuclei were stained with DAPI for 5 min at room temperature. Cell fluorescence were captured under the confocal microscope at ×40 magnification (Carl Zeiss AG).

HT29 and HCT116 cells were inoculated into 96-well plates (1×10³ cells/well) and cultured at 37°C. At indicated time points (0, 24, 48 and 72 h), HT29 and HCT116 cells in 96-well plates were further incubated with 10 µl CCK-8 reagent (Sigma-Aldrich; Merck KGaA) according to the manufacturer's instructions for 2 h at 37°C. Cell viability was calculated based on the absorbance at 450 nm.

HT29 and HCT116 cells were made into a cell suspension and inoculated into 6-well plates (200 cells/well). After culturing for 14 days, HT29 and HCT116 cells were fixed using 4% paraformaldehyde (Sigma-Aldrich; Merck KGaA) for 15 min at room temperature and stained with 0.1% crystal violet (Sigma-Aldrich; Merck KGaA) for 15 min at room temperature. The number of colonies (>50 cells) was assessed manually.

Transwell inserts with 8 µm pore size (Corning, Inc.) were inserted into 6-well plates. HT29 and HCT116 cells (1×10⁵ cells/well) were inoculated into the upper chamber of Transwell and cultured in serum-free medium (Gibco; Thermo Fisher Scientific, Inc.). For cell invasion assays, the upper chamber of Transwell was covered with 50 µl Matrigel (BD Biosciences) at 37°C for 2 h, which was the only difference to the migration assay. Complete medium supplemented with 10% FBS (Gibco; Thermo Fisher Scientific, Inc.) was added to the lower chamber of Transwell. After 24 h of cultivation, cells were fixed with 4% paraformaldehyde (Sigma-Aldrich; Merck KGaA) for 20 min at room temperature and stained with 0.1% crystal violet (Sigma-Aldrich; Merck KGaA) for 10 min at room temperature. The number of migrated and invaded cells was determined under a light microscope (Zeiss AG) in five random fields (magnification, ×100).

The circInteractome database (26) (https://circinteractome.nia.nih.gov/) was screened to identify miRNAs that were adsorbed by circ-FOXO3. miRanda (27) (http://www.microrna.org/) was used to search the target genes of miR-543.

The wild-type (WT) and mutant type (MUT) fragments of circ-FOXO3, and WT and MUT type 3′-untranslated region (3′-UTR) of LATS1 were synthesized and constructed into the pGL3 vector (Promega Corporation). Then, 100 nM circ-FOXO3 WT, circ-FOXO3 MUT, LATS1 WT 3′-UTR, LATS1 MUT 3′-UTR, miR-NC and miR-543 mimic were transfected into HT29 and HCT116 cells using Lipofectamine^® 2000 (Thermo Fisher Scientific, Inc.). After 48 h of transfection, the luciferase activity was determined using a microplate reader (Thermo Fisher Scientific, Inc.) and normalized to Renilla luciferase activity.

Biotinylated-miR-543 probe and biotinylated NC (Bio-NC) probe were generated by Guangzhou RiboBio Co., Ltd. The RNA pull-down was performed as previously described (28). To produce probe-coated beads, miR-543 (Guangzhou RiboBio Co., Ltd.) was incubated with C-1 magnetic beads (Thermo Fisher Scientific, Inc.) for 2 h at 25°C. The cell lysates were then harvested and treated with the 50 pmol miR-543 probe or Bio-NC probe overnight at 4°C. After washing with wash/binding buffer, the RNA complexes adsorbed in the beads were collected and used to conduct RT-qPCR assay and northern blot analysis. For northern blot analysis, 30 µg RNAs were separated in a 1% agarose-formaldehyde gel and transferred to Hybond-N+ membrane (Beyotime Institute of Biotechnology). Then, the membranes were hybridized with digoxin-labeled DNA oligonucleotides specific to circ-FOXO3 (Guangzhou RiboBio Co., Ltd.) at 37°C. The membranes were exposed to phosphorimager screens and analyzed using Image Lab V3.0 software (Bio-Rad Laboratories, Inc.).

Proteins were isolated from HT29 and HCT116 cells using RIPA buffer (Thermo Fisher Scientific, Inc.) and quantified using the BCA method. The extracted proteins (40 µg) were then separated on 10% SDS-PAGE and transferred to a PVDF membrane (Thermo Fisher Scientific, Inc.). After blocking with 4% skimmed milk for 1 h at room temperature, the membrane was probed with rabbit anti-human polyclonal LATS1 antibody (1:500; cat. no. ab70562), mouse anti-human monoclonal E-cadherin antibody (1:500; cat. no. ab76055), rabbit anti-human polyclonal N-cadherin antibody (1:500; cat. no. ab76057), rabbit anti-human polyclonal Vimentin antibody (1:500; cat. no. ab137321), mouse anti-human monoclonal MMP9 antibody (1:500; cat. no. ab119906) and rabbit anti-human polyclonal GAPDH antibody (1:1,000; cat. no. ab9485; all Abcam) at 4°C for 12 h. After the washing steps, the membrane was incubated with HRP-conjugated rabbit polyclonal anti-human IgG H&L (1:1,000; cat. no. ab6759) or HRP-conjugated mouse monoclonal anti-human IgG H&L (1:1,000; cat. no. ab436; both Abcam) at room temperature for 1 h. GAPDH was used as the loading control. The blots were visualized using ECL western blotting substrate (Thermo Fisher Scientific, Inc.).

SPSS Statistics 22.0 (IBM Corp.) was utilized to conduct data analysis. Data are presented as the mean ± SD. Student's t-test or one-way ANOVA with Tukey's post hoc test were used for analyzing the significance between different groups. A paired t-test was used to compare the difference between tumor tissues and adjacent tissues. The relationship between circ-FOXO3 expression and the clinicopathological parameters of patients with CRC was determined using a χ² test. The overall survival of patients with CRC was determined using Kaplan-Meier analysis with the log-rank test. The correlations among the expression levels of circ-FOXO3, miR-543 and LATS1 were determined via correlation analysis with a Pearson test. All experiments were repeated at least three times. P<0.05 was considered to indicate a statistically significant difference.

To assess the effects of circ-FOXO3 on CRC, the aberrant expression of circ-FOXO3 in CRC tissues was investigated. A significant downregulation of circ-FOXO3 expression was identified in tumor tissues compared with healthy tissues of 70 patients with CRC (P<0.001; Fig. 1A). In addition, circ-FOXO3 expression in CRC cells, including HT29, HCT116, HCT8, LOVO, SW480 and SW620, was significantly lower compared with wild-type colon epithelial cell line (all P<0.01; Fig. 1B).

To determine the association between circ-FOXO3 and clinical characteristics of patients with CRC, 70 patients with CRC were divided into the circ-FOXO3 high expression group and circ-FOXO3 low expression group based on the median value of circ-FOXO3 expression in CRC tissues as the cut-off value. Kaplan-Meier analysis demonstrated that the low expression of circ-FOXO3 was positively associated with poor overall survival of patients with CRC (P=0.0356; Fig. 1C). In addition, circ-FOXO3 expression was significantly associated with tumor size (P<0.01), distant metastasis (P=0.0003), differentiation (P=0.0002), lymph node metastasis (P<0.0001) and TMN stages (P=0.0081) (Table I). Therefore, these results suggested that circ-FOXO3 was downregulated in CRC, and associated with the overall survival, tumor size, distant metastasis, differentiation, lymph node metastasis and TMN stages of patients with CRC.

To investigate the effect of circ-FOXO3 on CRC, the role of circ-FOXO3 in CRC cell features was determined. It was identified that HT29 and HCT116 cells presented the lowest circ-FOXO3 expression compared with that in HCT8, LOVO, SW480 and SW620 cells (Fig. 1B). Thus, HT29 and HCT116 cells were used for follow-up experiments. First, a RNase R digestion assay was conducted in HT29 and HCT116 cells to assess the circular characteristics of circ-FOXO3. circ-FOXO3 was resistant to RNase R digestion, which indicated that circ-FOXO3 was a circRNA (P>0.05; Fig. 2A).

Subsequently, HT29 and HCT116 cells were transfected with pcDNA circ-FOXO3 plasmid to overexpress circ-FOXO3 expression, which were evaluated using RT-qPCR (P<0.0001 and P<0.001; Fig. 2B). The effect of circ-FOXO3 overexpression on CRC cell aggressive features was investigated after circ-FOXO3 overexpression. circ-FOXO3 overexpression significantly suppressed the viability of HT29 and HCT116 cells (all P<0.0001; Fig. 2C). Colony formation results indicated that the colony numbers of HT29 and HCT116 cells were significantly reduced after overexpression of circ-FOXO3 (all P<0.001; Fig. 2D). In addition, Transwell assays results found that circ-FOXO3 overexpression significantly inhibited cell migratory and invasive abilities (all P<0.0001; Fig. 2E and F). circ-FOXO3 overexpression also markedly increased the expression of E-cadherin, and notably decreased the expression levels of N-cadherin, Vimentin and MMP9 (Fig. 2G). Thus, these findings indicated that overexpression of circ-FOXO3 suppressed CRC cell proliferation, migration and invasion.

According to the aforementioned results, circ-FOXO3 presented aberrant expression in CRC and inhibited CRC cell aggressive features, but the underlying mechanism of their action remains to be determined. To identify the regulatory mechanism of circ-FOXO3 on CRC cell features, the location of circ-FOXO3 in CRC cells was first examined. A significantly higher expression of circ-FOXO3 was found in the cytoplasm compared with the nucleus of HT29 and HCT116 cells (all P<0.0001; Fig. 3A), which suggested that circ-FOXO3 was mostly located in the cytoplasm. circRNA in cytoplasm usually functions as a miRNA sponge (29). Therefore, it was hypothesized that circ-FOXO3 may function as a miRNA sponge in CRC cells to regulated CRC cell aggressive features.

The circInteractome database (https://circinteractome.nia.nih.gov/) was screened to identify miRNAs that are adsorbed by circ-FOXO3. A miR-543 binding site was found in circ-FOXO3, and the predicted sequences of circ-FOXO3 and miR-543 are presented in Fig. 3B. The transfection efficiency of miR-543 mimic was demonstrated using RT-qPCR (all P<0.0001; Fig. S1A). Luciferase assay results demonstrated that miR-543 overexpression inhibited the relative luciferase activity in cells transfected with WT circ-FOXO3 (all P<0.001), while no significant effect was found in cells transfected with MUT circ-FOXO3 (Fig. 3B). RNA pull-down results indicated a significantly higher level of enrichment of circ-FOXO3 using the biotinylated-miR-543 probe compared with biotinylated-NC group in HT29 and HCT116 cells (all P<0.01; Fig. 3C). Moreover, overexpression of circ-FOXO3 decreased miR-543 expression (all P<0.001; Fig. 3D).

To further examine the association between miR-543 and circ-FOXO3, miR-543 expression in patients with CRC was detected, and it was found that miR-543 expression in tumor tissues was significantly upregulated compared with healthy tissues (P<0.0001; Fig. 3E). Pearson correlation analysis demonstrated a moderate negative correlation between the expression levels of miR-543 and circ-FOXO3 in CRC (P<0.0001; R=−0.5955; Fig. 3F). Thus, circ-FOXO3 may function as a miR-543 sponge.

Next, miR-543 targets were screened using miRanda (http://www.microrna.org/). LATS1 was considered as a potential target of miR-543 and the putative binding site sequences are presented in Fig. 4A. Luciferase assay was conducted to identify whether miR-543 targeted LATS1. It was found that miR-543 overexpression suppressed relative luciferase activity in cells transfected with WT LATS1 (all P<0.001), while no significant effect was observed in cells transfected with mutant type LATS1 (Fig. 4A).

The transfection efficiency of miR-543 mimic and inhibitor in HT29 and HCT116 cells were determined via cell fluorescence, which indicated that the miR-543 mimic notably increased miR-543 expression, while miR-543 inhibitor decreased miR-543 expression (Fig. 4B). In addition, the transfection efficiency of the miR-543 inhibitor was verified using RT-qPCR (all P<0.001; Fig. S1B). The miR-543 inhibitor elevated the mRNA expression of LATS1 in HT29 (P<0.001) and HCT116 (P<0.001) cells, while miR-543 mimic inhibited the mRNA expression of LATS1 in HT29 (P<0.001) and HCT116 (P<0.01) cells (Fig. 4C). Similarly, the protein expression of LATS1 oppositely modulated by miR-543 (Fig. 4D). The miR-543 mimic decreased circ-FOXO3 mRNA expression (all P<0.01), while miR-543 inhibitor significantly increased the expression of circ-FOXO3 (P<0.01 and P<0.001; Fig. 4E). Furthermore, the inhibitory effect of miR-543 mimic on LATS1 was reversed by circ-FOXO3 overexpression in HT29 and HCT116 cells (all P<0.01; Fig. 4F).

To further evaluate the association between LATS1 and circ-FOXO3, the expression of LATS1 was detected in patients with CRC, and it was identified that LATS1 was significantly downregulated in CRC tumor tissues compared with healthy tissues (P<0.0001; Fig. 4G). In addition, Pearson correlation analysis demonstrated a weak positive correlation between the expression levels of circ-FOXO3 and LATS1 (P<0.0001; R=0.3239; Fig. 4H). Collectively, these results suggested that circ-FOXO3 increased LATS1 expression via sponging miR-543.

Based on the aforementioned results, it was suggested that the effect of circ-FOXO3 on CRC aggressive features may be mediated via miR-543 and LATS1. The transfection efficiency of si-LATS1 was verified using RT-qPCR (all P<0.001; Fig. S1C). The inhibitory role of circ-FOXO3 overexpression on cell viability was abolished by both miR-543 overexpression and LATS1 knockdown in HT29 and HCT116 cells (all P<0.001; Fig. 5A). Colony formation assay results demonstrated that both miR-543 overexpression (all P<0.01) and LATS1 knockdown (P<0.01 and P<0.05) reversed the inhibitory effect of circ-FOXO3 overexpression on colony formation ability in HT29 and HCT116 cells (Fig. 5B). Furthermore, Transwell assay results identified that the inhibited migratory and invasive abilities exerted by circ-FOXO3 overexpression were blocked by miR-543 overexpression (P<0.05-P<0.001) or LATS1 knockdown (P<0.01 and P<0.05) in HT29 and HCT116 cells (Fig. 5C and D). The promotion effect of circ-FOXO3 overexpression on E-cadherin expression, and its inhibitory effects on N-cadherin expression were also markedly reversed by miR-543 overexpression or LATS1 knockdown in HT29 and HCT116 cells (Fig. 5E). Therefore, these results indicated that both miR-543 overexpression and LATS1 knockdown inhibited circ-FOXO3-induced attenuated CRC aggressive features.