CircTADA2A suppresses the progression of colorectal cancer via miR-374a-3p/KLF14 axis

Seventy pairs of CRC specimens and matching paracancerous specimens were obtained from CRC patients at the First Affiliated Hospital of Zhengzhou University. All tissues were stored in liquid nitrogen for the detection of circTADA2A, miR-374a-3p and KLF14. All tissue samples were pathologically diagnosed. The study was authorized by the Ethic Committee of the First Affiliated Hospital of Zhengzhou University. Written informed consent was provided from each subject.

Human CRC cell lines (HCT116, SW620, LoVo and SW480) and normal human colon epithelial cell line NCM460 were obtained from BeNa Culture Collection (Beijing, China). These cell lines were maintained in Dulbecco’s Modified Eagle Medium (DMEM; Gibco, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS; Gibco), 10% penicillin (100 U/mL) and 10% streptomycin (100 μg/mL) in an incubator with the environment of 37 °C and 5% CO₂.

RNA was isolated with Trizol reagent (Life Technologies, Carlsbad, CA, USA). The purity and the concentration of different RNA samples were measured using NanoDrop ND-1000. The synthesis of complementary DNA (cDNA) was performed with Geneseed® II First Strand cDNA Synthesis Kit (Geneseed, Guangzhou, China) and All-in-OneTM miRNA First stand cDNA Synthesis Kit (GeneCopoeia, Rockville, MD, USA). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and U6 served as the internal references in this study. The quantification of circTADA2A, transcriptional adaptor 2A (TADA2A) messenger RNA (mRNA), miR-374a-3p and KLF14 was carried out using the 2^−ΔΔCt method. The primers were shown as below: circTADA2A (Forward, 5′-TGTGCACCAAGACCAAGGAG-3′; Reverse, 5′-AGGAAAATCTGAAGTAGTGA-3′), TADA2A (Forward, 5′-CCTTTTTTCCTCTGCTTGCA-3′; Reverse, 5′-ATCCTGCCAATTTCCAAAGC-3′), miR-374a-3p (Forward, 5′-CUUAUCAGAUUGUAUUGUAAUU-3′; Reverse, 5′-AAUUACAAUACAAUCUGAUAAG-3′), KLF14 (Forward, 5′-TCGGAGGTGGGTGCGGCGCC-3′; Reverse, 5′-GGAGCCCTCGCCAGAGCTGC-3′), U6 (Forward, 5′-CTCGCTTCGGCAGCACA-3′; Reverse, 5′-AACGCTTCACGAATTTGCGT-3′), GAPDH (Forward, 5′-GGAGCCAAAAGGGTCATC-3′; Reverse, 5′-CCAGTGAGTTTCCCGTTC-3′), 18S rRNA (Forward, 5′-ACGGGCGCTGACCCCCCTTC-3′; Reverse, 5′-AGGGCAGACGTTCGAATGGG-3′).

Total RNA samples were treated with 3 U/mg RNase R (Epicentre Technologies, Madison, WI, USA) for 30 min at room temperature. qRT-PCR was employed to detect the levels of circTADA2A and TADA2A mRNA. Subcellular fractionation was conducted with PARIS™ Kit (Invitrogen, Carlsbad, CA, USA). 18S rRNA served as the cytoplasmic marker while U6 acted as nuclear marker in this study.

CircTADA2A overexpression plasmid (circTADA2A) and control empty vector (vector), KLF14 small interfering RNA (si-KLF14) and siRNA control (si-con), miR-374a-3p mimic (miR-374a-3p) and control (miR-con), miR-374a-3p inhibitor (anti-miR-374a-3p) and control (anti-miR-con) were purchased from Genepharma (Shanghai, China), and these plasmids or RNAs were transfected into CRC cells using Lipofectamine 3000 (Invitrogen).

BALB/c nude mice (5 weeks old) were purchased from Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China) and randomly divided into 3 groups (n = 6). HCT116 or LoVo cells were stably transfected with vector or circTADA2A. The back region of BALB/c mice was subcutaneously injected with CRC cells stably expressing vector or circTADA2A or un-transfected CRC cells. The volume of tumors was measured every 7 d after inoculation for 7 d using a vernier caliper with the method of volume = π × (length × width²)/6. The mice were killed after inoculation for 35 d, and the tumors were weighed. Tumor tissues were subjected to measure the expression of circTADA2A by qRT-PCR. The procedures in this study were permitted by the Animal Research Committee of the First Affiliated Hospital of Zhengzhou University.

The miR-374a-3p binding sequence in circTADA2A or the 3’untranslated region (3’UTR) of KLF14 was amplified and inserted to pGL3 luciferase reporter vector (Promega, Madison, WI, USA), and the recombinant luciferase reporter vector was termed as circTADA2A-wild-type (circTADA2A-WT) or KLF14-WT. The matching mutant type binding sites with miR-374a-3p in circTADA2A or the 3’UTR of KLF14 were also cloned to pGL3 luciferase reporter vector, generating circTADA2A-MUT or KLF14-MUT. LoVo and HCT116 cells were transfected with miR-con or miR-374a-3p and the above constructed reporter plasmids. The luciferase activities in different groups were determined by the dual-luciferase reporter assay system (Promega).

Magna RIP™ RNA-Binding Protein Immunoprecipitation kit (Millipore, Billerica, MA, USA) was used to conduct RIP assay. CRC cells were lysed with RIP buffer (Millipore). The cell lysate was incubated with sepharose beads (Bio-Rad, Hercules, CA, USA) pre-coated with Argonaute-2 antibody (Anti-Ago2). Immunoglobulin G antibody (Anti-IgG) served as the control in this study. qRT-PCR was applied to measure the abundance of circTADA2A, miR-374a-3p and KLF14 in the immunoprecipitated complexes.

CRC cells were harvested with cold phosphate buffer saline (PBS), and fixed in 70% cold ethanol solution (diluted with PBS) at − 20 °C overnight. After incubating with RNase (10 μM) for 30 min, propidine iodide (PI; 20 mg/mL; Solarbio, Beijing, China) was used to stain CRC cells for 20 min at 37 °C. The cell cycle was evaluated on a flow cytometer (BD Biosciences, San Jose, CA, USA).

CRC cells were collected and rinsed using cold PBS. Subsequently, CRC cells were suspended in Annexin-V binding buffer, and then these CRC cells were dyed using Annexin V combined fluorescein isothiocyanate (Annexin V-FITC; Solarbio) and PI (Solarbio) simultaneously in the dark for 15 min. The apoptotic CRC cells were tested on a flow cytometer (BD Biosciences).

Seahorse XF^e 96 Extracellular Flux Analyzer (Seahorse Bioscience, Billerica, MA, USA) was used to detect the extracellular acidification rate (ECAR) and cellular oxygen consumption rate (OCR) using Seahorse XF Glycolysis Stress/Cell Mito Stress Test Kit (Seahorse Bioscience). For the detection of ECAR, Glucose, Oligomycin and 2-deoxyglucose (2-DG) were sequentially added to the wells of Seahorse XF 96 cell culture microplate. For the measure of OCR, Oligomycin, p-trifluoromethoxy carbonyl cyanide phenylhydrazone (FCCP) and the mitochondrial complex I inhibitor rotenone plus the mitochondrial complex III inhibitor antimycin A (Rote+AA) were sequentially added.

Glucose Uptake Colorimetric Assay kit (Biovision, Milpitas, California, USA), Lactate Assay Kit II (Biovision) and ATP Colorimetric Assay kit (Biovision) were utilized to detect the glucose consumption and the production of lactate and ATP in CRC cells according to the manufacturer’s instructions.

After relevant treatment, CRC cells were harvested using PBS and disrupted using Radioimmunoprecipitation assay (RIPA) solution (Beyotime, Shanghai, China). Protein samples (30 μg) were run on sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gel and transferred to the polyvinylidene fluoride (PVDF) membrane. The PVDF membrane was blocked using 5% non-fat milk for 1 h, and then incubated with anti-KLF14 (ab85476; Abcam, Cambridge, MA, USA) or anti-β-actin (ab8226, Abcam) at 4 °C overnight. After washing with PBS-Tween 20 (PBST) for 3 times, the membrane was incubated with the secondary antibody (ab205718) for 2 h. After washing with PBST, the protein bands were measured by the enhanced chemiluminescent (ECL) system (Beyotime).

GraphPad Prism 7.0 was used to analyze the data and generate the graphs. Student’s t-test or one-way analysis of variance (ANOVA) followed by Tukey’s test were used to calculate the P value between two groups or more than two groups. The correlation among the expression of miR-374a-3p, circTADA2A and KLF14 was analyzed using Spearman’s correlation coefficient. The differences between the enrichment of circTADA2A and the clinical characteristics of CRC patients were analyzed using χ² test. P < 0.05 was identified as statistically significant.

According to the data of circRNA microarray from GEO database (GSE126094), we analyzed the expression patterns of circRNAs in 10 pairs of CRC tissues and non-tumor specimens. The top 20 differentially expressed circRNAs between CRC samples and adjacent normal samples were shown in Fig. 1a. The expression difference degree of these 20 circRNAs in CRC tissues and non-tumor specimens was listed in Table 1. Among these circRNAs, circTADA2A was the most down-regulated circRNA in CRC specimens (Table 1), thus circTADA2A was chosen for the following study. In addition, we also measured the expression of circTADA2A in a total of 70 pairs of CRC tissues and paracancerous tissues. A significant decrease in the expression of circTADA2A was observed in CRC tissues relative to normal tissues (Fig. 1b). The association between the expression of circTADA2A and the clinicopathologic features of CRC patients was shown in Table 2. CircTADA2A expression was negatively correlated with the tumor size, pT stage and distant metastasis of CRC tumors (Table 2). A panel of CRC cell lines, including HCT116, SW620, LoVo and SW480 and normal human colon epithelial cell line NCM460 were used to explore the expression profile of circTADA2A. As shown in Fig. 1c, circTADA2A was markedly down-regulated in CRC cells than that in NCM460 cells, especially in HCT116 and LoVo cells. Taken together, circTADA2A was notably down-regulated in CRC.

To test the circular structure and the stability of circTADA2A, Oligo (dT)₁₈ primers and RNase R were used. As displayed in Fig. 2a and b, the level of circTADA2A was low in Oligo (dT)₁₈ primers group relative to Random primers group, while the expression of its linear form remained unchanged in these two groups, suggesting that circTADA2A had no poly(A) tail. Besides, as exhibited in Fig. 2c and d, circTADA2A was resistant to RNase R compared with TADA2A mRNA. To explore the subcellular localization of circTADA2A, we isolated the nuclear fraction of CRC cells from the cytoplasmic fraction using PARIS™ Kit. CircTADA2A mainly distributed in the cytoplasm of CRC cells (Fig. 2e and f). These results revealed that circTADA2A possessed the circular structure and it mainly localized in the cytoplasm of CRC cells.

Xenograft tumor model was established to explore the function of circTADA2A in vivo. The mean volume and weight of CRC tumors derived from LoVo cells or HCT116 cells overexpressing circTADA2A were lower relative to Mock group or vector group (Fig. 3-3a-d). The representative tumor images in different groups were shown in Fig. 3e and f. The overexpression efficiency was also verified by qRT-PCR. There was a significant up-regulation in the abundance of circTADA2A in circTADA2A group than that in Mock or vector group (Fig. 3g and h). These findings indicated that circTADA2A suppressed the growth of CRC tumors in vivo.

Circbank and starbase softwares were utilized to seek the targets of circTADA2A (Sup Table 1 and Sup Table 2). As shown in Fig. 4a, there were 28 miRNAs that were predicted by both the two bioinformatic softwares. According to the starbase v3.0 project, among these 28 miRNAs, there existed 7 miRNAs that were highly expressed in CRC tissues in contrast to that in normal tissues, while a total of 11 miRNAs were found to be down-regulated in CRC tumor tissues relative to normal tissues (Table 3). In view of the low expression of circTADA2A in CRC tissues, we concentrated on these 7 highly expressed miRNAs. The function of miR-374a-3p had not been reported, so we focused on miR-374a-3p. As indicated in Fig. 4b, miR-374a-3p was abnormally up-regulated in colon carcinoma tissues (n = 450) relative to normal tissues (n = 8) according to the data from starbase v3.0 project. A marked up-regulation in the level of miR-374a-3p was also observed in CRC tissues (n = 70) compared with that in matching normal tissues (n = 70) (Fig. 4c). In addition, miR-374a-3p showed high expression in CRC cells than that in NCM460 cells (Fig. 4d). There was a negative correlation between the expression of circTADA2A and miR-374a-3p with the P value of 0.007 and r value of − 0.3198 (Fig. 4e). To confirm the target relationship between miR-374a-3p and circTADA2A, we mutated the binding sites with miR-374a-3p in circTADA2A (Fig. 4f). The sequence of circTADA2A containing wild-type or mutant type binding sites with miR-374a-3p was inserted into pGL3 luciferase reporter plasmid, generating circTADA2A-WT or circTADA2A-MUT. LoVo and HCT116 cells were co-transfected with miR-con or miR-374a-3p and the forementioned luciferase reporter plasmids. As indicated in Fig. 4g and h, the luciferase activity was dramatically reduced in circTADA2A-WT group when co-transfected with miR-374a-3p instead of miR-con, while the luciferase activity remained unaffected in circTADA2A-MUT group when co-transfected with miR-con or miR-374a-3p. Ago2 is an important component of RNA-induced silencing complex (RISC) that contained miRNAs, thus we tested if there was spatial target relationship between miR-374a-3p and circTADA2A in RISC via using Ago2 antibody (Anti-Ago2). CircTADA2A and miR-374a-3p were all enriched in Anti-Ago2 group compared with Anti-IgG group, suggested that miR-374a-3p bound to circTADA2A in CRC cells (Fig. 4i and j). Taken together, miR-374a-3p was a direct target of circTADA2A in CRC cells.

To address whether circTADA2A exerted its role through targeting miR-374a-3p, we conducted rescue experiments. LoVo and HCT116 cells were transfected with the following four groups: vector, circTADA2A, circTADA2A + miR-con or circTADA2A + miR-374a-3p. The transfection efficiencies of circTADA2A and miR-374a-3p were both high in the two CRC cell lines (Fig. 5a and b). CircTADA2A overexpression elevated the percentage of CRC cells in G0-G1 stage, suggesting the inhibition of cell cycle in circTADA2A group, while the introduction of miR-374a-3p counteracted the inhibitory effect of circTADA2A accumulation on the cycle of CRC cells (Fig. 5c and d, Supplementary 1A and 1B). The apoptotic rate was notably increased in circTADA2A group, and the addition of miR-374a-3p impeded the apoptosis of CRC cells (Fig. 5e and f, Supplementary 1C and 1D). Warburg effect is one of the hallmarks of cancers, characterized by the promotion of glycolysis and the inhibition of the oxidative phosphorylation with the presence of oxygen [19]. We explored the influence of circTADA2A/miR-374a-3p axis on the Warburg effect of CRC cells through measuring the ECAR, OCR, glucose uptake, lactate production and ATP production. As exhibited in Fig. 5-5g-p, circTADA2A accumulation resulted in the inhibition of the ECAR, glucose uptake and the production of lactate and ATP and the promotion on the OCR of CRC cells, and these effects were attenuated by the co-transfection of circTADA2A and miR-374a-3p. These findings indicated that circTADA2A suppressed the cell cycle and glycolysis and promoted the apoptosis of CRC cells at least partly through targeting miR-374a-3p.

KLF14 was predicted as a target of miR-374a-3p by DIANA TOOL (Fig. 6a). The luciferase activity was decreased in KLF14-WT and miR-374a-3p co-transfected group other than miR-con and KLF14-WT co-transfected group, and there was no significant change in the luciferase activity of KLF14-MUT group with the addition of miR-con or miR-374a-3p (Fig. 6b and c). In addition, the levels of KLF14 and miR-374a-3p were high in the immunoprecipitated complex pulled-down by Ago2 antibody, suggested that KLF14 could bind to RNA induced silencing complex (RISC) (Fig. 6d and e). The enrichment of KLF14 mRNA and protein was notably reduced in CRC tissues relative to normal tissues (Fig. 6f and g). The mRNA and protein of KLF14 were also down-regulated in four CRC cell lines compared with NCM460 cells (Fig. 6h and i). MiR-374a-3p overexpression significantly down-regulated the protein expression of KLF14 in both CRC cell lines (HCT116 and LoVo) and NCM460 cell line (Fig. 6j). Collectively, KLF14 was a target of miR-374a-3p in CRC cells.

The transfection efficiencies of anti-miR-374a-3p and si-KLF14 were confirmed by qRT-PCR and Western blot. As shown in Fig. 7a, the transfection of anti-miR-374a-3p prominently reduced the abundance of miR-374a-3p in LoVo and HCT116 cells. The protein expression of KLF14 was notably declined in si-KLF14 group than that in si-con group (Fig. 7b). MiR-374a-3p expression was notably decreased with the transfection of anti-miR-374a-3p, and the addition of si-KLF14 did not affect the abundance of miR-374a-3p in CRC cells (Supplementary Fig. 2A and 2B). Besides, miR-374a-3p silencing up-regulated KLF14 expression in CRC cells, and the addition of si-KLF14 partly counteracted anti-miR-374a-3p-mediated promoting effect on KLF14 protein expression (Supplementary Fig. 2C and 2D). The cell cycle of CRC cells was blocked and the apoptosis was triggered with the depletion of miR-374a-3p, and the simultaneous transfection of anti-miR-374a-3p and si-KLF14 reversed the influence on the cell cycle and apoptosis caused by miR-374a-3p intervention (Fig. 7-7c and f). Furthermore, the interference of KLF14 also overturned the suppressive effects of miR-374a-3p silencing on the ECAR, glucose consumption, lactate production and ATP generation and the promoting effect on the OCR of CRC cells (Fig. 7-7g-p). Collectively, KLF14 depletion could partly reverse the biological effects of CRC cells mediated by miR-374a-3p intervention.

We explored the linear relationship between KLF14 mRNA and circTADA2A or miR-374a-3p in CRC tissues (n = 70) through using Spearman’s correlation coefficient. KLF14 mRNA was positively correlated with circTADA2A and inversely correlated with miR-374a-3p (Fig. 8a and b). To clarify the relationship among KLF14, miR-374a-3p and circTADA2A, LoVo and HCT116 cells were transfected with the following four groups: vector, circTADA2A, circTADA2A + miR-con or circTADA2A + miR-374a-3p. The protein expression of KLF14 was up-regulated with the overexpression of circTADA2A, and the co-transfection of miR-374a-3p and circTADA2A declined the protein expression of KLF14 (Fig. 8c). Moreover, LoVo and HCT116 cells were co-transfected with si-circTADA2A and anti-miR-374a-3p. The depletion of miR-374a-3p recovered the protein abundance of KLF14, which was down-regulated by the transfection of si-circTADA2A (Fig. 8d). Based on these results, we concluded that circTADA2A could up-regulate KLF14 through functioning as a ceRNA of miR-374a-3p.

Based on the above findings, circTADA2A served as a tumor suppressor and it was down-regulated in CRC. The ectopic expression of circTADA2A could restrain the tumor growth, cell cycle and glycolysis and promote the apoptosis of CRC cells through miR-374a-3p/KLF14 axis (Fig. 9).