RNA sequencing reveals the expression profiles of circRNA and indicates that circDDX17 acts as a tumor suppressor in colorectal cancer

A total of 60 pairs of CRC and adjacent normal mucosa tissues were collected from patients undergone surgery between September 2016 and February 2018 at the Department of General Surgery, Beijing Chao-Yang Hospital, Capital Medical University (Beijing, China). The specimens were immediately frozen in liquid nitrogen after collection and stored at − 80 °C until RNA extraction. Among the 60 pairs of CRC tissues, 4 pairs were selected for RNA-seq, 20 pairs (not including the tissues for RNA-seq) were used for qRT-PCR to validate the RNA-seq data, and all the 60 pairs were used for qRT-PCR to detect the level of circDDX17. All of the specimens were verified by histopathology. None of these patients received preoperative chemotherapy, radiotherapy, or targeted therapy.

The human CRC cell lines SW480, SW620, HT29, LoVo, HCT116, and RKO were purchased from the American Type Culture Collection (Manassas, VA, USA). Cells were cultured in Dulbecco’s Modified Essential Medium (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, NY, USA), 100 U/ml penicillin and 100 μg/ml streptomycin (Gibco, NY, USA) at 37 °C in 5% humidified CO₂ atmosphere.

Total RNAs were extracted with Trizol (Invitrogen, Carlsbad, CA, USA). RNA integrity was analyzed with Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). RNA concentration was tested by Qubit RNA Assay Kit in Qubit Fluorometer (Invitrogen, Carlsbad, CA, USA). Total RNA samples that meet the following requirements were used in subsequent experiments: RNA integrity number (RIN) ≥ 7.0 and a 28S:18S ratio ≥ 1.5.

Sequencing libraries were generated and sequenced by CapitalBio Technology (Beijing, China). A total amount of 5 μg RNA per sample was used. Briefly, total RNAs were subjected to ribosomal RNA (rRNA) removal using the Ribo-Zero™ Magnetic Kit (Epicentre Technologies, Madison, WI, USA). To remove linear RNAs, total RNAs were digested with RNase R (Epicentre Technologies, Madison, WI, USA). The NEBNext Ultra RNA Library Prep Kit for Illumina (NEB, USA) was used to construct the libraries for sequencing according to the manufacturer’s protocol. The RNA was then fragmented into pieces of ~ 300 base pairs (bp) in length in NEBNext First Strand Synthesis Reaction Buffer (5x). The first-strand cDNA was synthesized from the RNA fragments by reverse transcriptase and random hexamer primers, and the second-strand cDNA was synthesized in Second Strand Synthesis Reaction Buffer with dUTP Mix (10x). The end of the cDNA fragment was subjected to an end repair process which included the addition of a single “A” base, followed by ligation of the adapters. After the ligation of Illumina sequencing adaptors, the second strand of cDNA was digested using the USER Enzyme (NEB, USA) to construct a chain specific library. To amplify the library DNA, libraries were purified and enriched by PCR. Then, the libraries were qualified by Agilent 2100 and quantified using KAPA Library Quantification kit (KAPA Biosystems, South Africa). Finally, the libraries were subjected to paired-end sequencing with pair end 150 bp reading length on an Illumina HiSeq sequencer (Illumina, San Diego, CA, USA).

FastQC software (v0.11.2) [14] was used to assess the sequencing quality of raw data of fastq format. Low quality data were filtered using NGSQC software (v2.3.2) [15]. The clean reads with high quality were then aligned to the human reference genome (GRCh38/hg38) using Tophat2 software (v2.0.13) [16] with default parameters. The sequencing data that cannot aligned to reference genome directly were subjected to the subsequent circRNA analysis by recognition of the reverse splicing event using Find_circ (v1.0) [17] and CIRCexplorer2 softwares [18]. Limma (v3.32.10) [19] package of R software was used to analyze the different expression of circRNAs. Function annotation and pathway enrichment analyses for the host genes of differentially expressed circRNAs were performed with KOBAS 3.0 [20].

Target miRNAs of circRNAs were predicted using miRanda software (v3.3a) [21]. Detailed setting parameters of miRanda software were as follows: Gap Open Penalty: − 9; Gap Extend Penalty: − 4; Score Threshold: 140; Energy Threshold: − 20 kcal/mol; Scaling Parameter: 4. The miRNAs associated with the KEGG pathway of COLORECTAL CANCER (Entry: map05210; https://​www.​kegg.​jp/​) were analyzed using DIANA-miRPath v.3 platform (Reverse Search module) [22]. Furthermore, we also screened target miRNAs of top 10 dysregulated circRNAs according to the KEGG pathway of MicroRNAs in cancer (section of “Colorectal cancer”; Entry: map05206; https://​www.​kegg.​jp/​), and target miRNAs associated with MicroRNAs in cancer pathway were identified. The maps of circRNA-miRNA interaction network were constructed using Cytoscape software [23].

Total RNAs from cells and tissue samples were isolated using Trizol (Invitrogen, Carlsbad, CA, USA). Specific divergent primers spanning the back-splice junction site of circRNAs were designed. To quantify the amount of mRNA and circRNA, cDNA was synthesized using PrimeScript™ RT reagent Kit (TaKaRa, Dalian, China). The real-time PCR analyses were performed using TB Green™ Premix Ex Taq™ II (TaKaRa, Dalian, China) and an ABI 7500 real-time PCR system (Applied Biosystems, Foster City, CA, USA). 18 s rRNA was employed as internal control. The relative expression of RNAs was calculated by 2^−ΔΔCt method. All the primers (listed in Additional file 1: Table S1) were synthesized by Sangon Biotech (Shanghai, China).

Small interfering RNAs (siRNAs) targeting the back-splice junction sites of circDDX17 and negative control oligonucleotide (Additional file 1: Table S2) were designed and synthesized by GenePharma (Shanghai, China). Transfections were performed with final concentrations of 50 nM of siRNAs using the Lipofectamine 3000 reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions.

For colony formation assay, 24 hours after transfection, SW480 and SW620 cells were seeded into 6-well plates and then incubated for about two weeks. Then the colonies were fixed with 4% paraformaldehyde for 15 min and stained with 0.1% crystal violet for 15 min at room temperature. The cell colonies were counted and photographed.

Cell proliferation assay was performed using CCK-8 kit (Dojindo Laboratories, Kumamoto, Japan) following the manufacturer’s instructions. Approximately 1 × 10³ transfected SW480 and SW620 cells in 100 μl were incubated in 96-well plates. At 0, 24, 48, 72 and 96 h, cells in each well were treated with the CCK-8 reagent (10 μl) and incubated at 37 °C for 2 h. The optical density (OD) at 450 nm was read using a Varioskan Flash Spectral Scanning Multimode Reader (Thermo Fisher Scientific, Waltham, MA, USA). Cells in each group were tested for 5 replicates.

Cell apoptosis was analyzed using the Annexin V-FITC Apoptosis Detection Kit (KeyGen Biotech, Nanjing, China) according to the manufacturer’s instructions. 48 h after transfection, SW480 and SW620 cells were stained with FITC and propidium iodide. Then flow cytometry was performed using a FACS Canto II (BD Biosciences) and the data were analyzed using BD FCSDiva Software and FCS Express 5 software (De Novo Software, Los Angeles, CA).

Cell migration and invasion assays were conducted using Transwell Permeable Supports, Polycarbonate (PC) Membrane (Corning, NY, USA), which was coated with (for invasion assays) or without (for migration assays) the matrigel (Corning, NY, USA) according to the manufacturer’s instructions. 24 h after transfection, cells in 200 μl serum-free medium were seeded into the upper compartment, with 700 μl culture medium containing 20% FBS in the lower compartment. After 24 h incubation, the cells located on the upper surfaces of the transwell chamber were erased with cotton swabs and the cells located on the lower surfaces were fixed with 4% paraformaldehyde for 15 min and stained with 0.1% crystal violet for 15 min at room temperature. The stained cells were photographed and counted in five randomly selected fields under a Leica DM4000B microscope (Leica, Wetzlar, Germany).

A total of 21,458 circRNAs were detected by RNA-Seq in 4 pairs of CRC and adjacent normal mucosa tissues. These circRNAs were widely distributed in all chromosomes including sex chromosomes X and Y (Fig. 1a). The scatter plot presented the difference in circRNA expression between the two groups (Fig. 1b). We evaluated Pearson’s correlation coefficient of significantly dysregulated circRNA expression level among different tissues, and the heatmap of inter-sample correlation (Fig. 1c) showed that there was an obvious difference of transcript expression levels between the CRC and control groups, but the difference was slight within each group. After screening of differentially expressed circRNAs by fold-change filtering (|log2(fold change)| > 1) and Student’s t-testing (p value < 0.05), 448 differentially expressed circRNAs were identified. Compared with the normal mucosa tissues, there were 394 significantly up-regulated and 54 significantly down-regulated circRNAs in CRC tissues. Volcano plots visualized the significantly differentially expressed circRNAs in CRC tissues (Fig. 1d). Most differentially expressed circRNAs derived from exons (Fig. 1e). 15 circRNAs were identified as new circRNAs that had not been annotated in the circBase or circ2Traits database [24, 25] (Additional files 2 and 3: Datas 1 and 2). The results of hierarchical clustering (Fig. 1f) suggested that the circRNA expression patterns were distinguishable between the CRC and control groups. The top 10 dysregulated circRNAs are listed in Table 1.

To verify the RNA-Seq data, qRT-PCR was performed for 5 up-regulated and 5 down-regulated circRNAs selected from the top 10 dysregulated circRNAs (two were the most dysregulated, and eight were randomly chosen). The expression levels of those 10 circRNAs were measured in another 20 pairs of CRC tissues. We designed specific divergent primers spanning the back-splice junction sites of circRNAs. The results showed that the expression of circASPHD1, circHOMER1, circAPLP2, circVAPA, and circTRAPPC9 was up-regulated, and the expression of circPRKAG2, circITFG2, circDDX17, circTRPM4, and circLMF1 was down-regulated (Fig. 2a). To verify the back-splice junction sequences, five circRNA qRT-PCR amplified products were sent to Sanger sequencing, and the sequencing results confirmed the back-splice junction sites (Fig. 2b). The qRT-PCR results were highly consistent with the RNA-seq data (Fig. 2c), confirming the reliability of the RNA-Seq results.

GO and KEGG analyses of the host genes of differentially expressed circRNAs were conducted to predict circRNA functions and molecular interactions among genes. GO analyses covered three domains: biological process, cellular component and molecular function. The top 10 enriched GO terms in biological process, cellular component, and molecular function are shown in Fig. 3a. The most significantly enriched GO terms in the biological process, cellular component, and molecular function categories were regulation of cell communication, autophagosome, and GTPase binding, respectively. The results of KEGG pathway analysis are presented in Fig. 3b. The host genes of differentially expressed circRNAs were mainly associated with superpathway of inositol phosphate compounds, FoxO signaling pathway, valine degradation, etc. Notably, the host genes of differentially expressed circRNAs were associated with important CRC-related pathways, such as Deleted in Colorectal Carcinoma (DCC) mediated attractive signaling, Netrin-1 signaling, Loss of Function of SMAD2/3 in Cancer, SMAD2/3 MH2 Domain Mutants in Cancer, TGFBR1 LBD Mutants in Cancer, SMAD4 MH2 Domain Mutants in Cancer, and Loss of Function of SMAD4 in Cancer.

Mounting studies have revealed that circRNAs can act as miRNA sponge molecules, which regulate the expression of genes by adsorbing miRNAs. Hence, we predicted the potential target miRNAs of differentially expressed circRNAs using miRanda software (v3.3a) (results for top 10 dysregulated circRNAs were presented in Additional file 4: Data 3). A circRNA-miRNA interaction network was constructed for the top 10 dysregulated circRNAs and their predicted binding miRNAs using Cytoscape software (Additional file 5: Figure S1). To further explore the functional roles of circRNA-miRNA interactions in CRC, we screened miRNAs associated with the KEGG pathway of COLORECTAL CANER using DIANA-miRPath v.3 platform (Reverse Search module; Additional file 6: Data 4). DIANA-miRPath v.3 is an online software suite designed to assess miRNAs regulating selected pathways based on the functions of experimentally supported miRNA target genes (using TarBase v7.0 method of miRPath software) or in silico predicted miRNA target genes (using microT-CDS v5.0 or TargetScan 6.2 method of miRPath software). In this study, three methods of miRPath software were combined to perform the analyses. Based on the analyses by combining the results from miRanda and miRPath softwares, a circRNA-miRNA interaction network potentially associated with the KEGG pathway of COLORECTAL CANCER was constructed using Cytoscape software (Fig. 4a). The network was based on the top 10 dysregulated circRNAs and target miRNAs with experimentally supported or in silico predicted target genes which were involved in the KEGG pathway of COLORECTAL CANCER, and the network showed that most miRNAs targeted by the top 10 dysregulated circRNAs were associated with the KEGG pathway of COLORECTAL CANCER (Additional file 4: Data 3). Furthermore, we also screened target miRNAs of top 10 dysregulated circRNAs according to the KEGG pathway of MicroRNAs in cancer, and target miRNAs associated with MicroRNAs in cancer pathway were listed in the Additional file 4: Data 3. In this study, we regarded up- or down-regulated circRNAs as potential oncogenes or tumor suppressors, respectively, and so it was with miRNAs in MicroRNAs in cancer pathway. Up-regulated (or down-regulated) circRNAs targeting down-regulated (or up-regulated) miRNAs in MicroRNAs in cancer pathway were identified as potential functional circRNAs associated with MicroRNAs in cancer pathway. Consequently, we found that among the top 10 dysregulated circRNAs, circPRKAG2, circITGAL, circDDX17, circASPHD1, circSPIDR, and circHOMER1 could potentially bind to miRNAs involved in the KEGG pathway of MicroRNAs in cancer (Fig. 4b). In summary, the above results collectively indicated that the top 10 dysregulated circRNAs might play important functional roles in the initiation and progression of CRC by interacting with miRNAs involved in the KEGG pathway of COLORECTAL CANCER or MicroRNAs in cancer.

As shown in Fig. 2a, circDDX17 was the most significantly differential expressed circRNA in qRT-PCR validation, and bioinformatic analyses indicated that it could potentially bind to hsa-miR-21-5p, which was involved both in the KEGG pathways of COLORECTAL CANCER and MicroRNAs in cancer (Additional file 4: Data 3; Fig. 4b). Hence, to further explore the functional roles of circRNA in CRC, we selected circDDX17 as a candidate circRNA for further investigation. CircDDX17 (hsa_circ_0002211) is derived from the DDX17 gene exon 2–8, with a spliced mature sequence length of 927 bp. We tested circDDX17 expression in 60 pairs of CRC tissues by qRT-PCR. The results showed that the expression of circDDX17 was significantly down-regulated in CRC tissues (Fig. 5a). 71.67% (43/60) of the 60 CRC patients showed decreased expression of circDDX17 in tumor tissues compared with matched adjacent normal mucosa tissues (Fig. 5b). Notably, we found that the expression of circDDX17 showed a decreased trend with advanced TNM stages(Fig. 5c). The expression level of circDDX17 in Stage III and IV patients was significantly lower than that of Stage I and II patients. In our study, stage I and II cases were grouped together due to the limited number of Stage I cases. Based on the qRT-PCR results, all the 60 patients were divided into high expression group and low expression group according to the median expression level (Fig. 5d). Statistical analyses showed that decreased expression of circDDX17 was significantly associated with lymphovascular invasion (p = 0.011), Depth of invasion (p = 0.007), lymph node metastasis (p = 0.020), distant metastasis (p = 0.012), and advanced TNM stage (p = 0.024) (Table 2). These results indicated that circDDX17 might play functional roles in the progression of CRC.

In order to investigate the biological functions of circDDX17 in CRC, we silenced the circDDX17 expression in CRC cell lines. To choose the CRC cell lines used for silencing of circDDX17, we measured the expression level of circDDX17 in six CRC cell lines. As shown in Fig. 6a, RKO cells showed the lowest expression of circDDX17; SW480 cells showed the highest expression, while in SW620, derived from a metastasis of the same tumor from which the SW480 was derived, the expression level was moderately decreased. Therefore, SW480 and SW620 cells were selected for silencing of circDDX17. We designed two siRNAs targeting the back-splice junction sites of circDDX17 to silence circDDX17 expression. As shown in Fig. 6b, the siRNAs significantly decreased circDDX17 expression level, but had no effect on its liner isoform. We chose si-circDDX17#1 for the subsequent experiments due to its higher efficiency of interference. Cell proliferation was measured by the CCK-8 assay (Fig. 6c), and silencing of circDDX17 significantly promoted cell proliferation in both SW480 and SW620 cells. The colony formation assay showed that knockdown of circDDX17 significantly increased colony-forming ability of SW480 and SW620 cells (Fig. 6d). In cell apoptosis assay, the si-circDDX17#1 cells exhibited significantly decreased apoptotic rate when compared with cells of negative control (Fig. 6e). Moreover, in the transwell migration and invasion assays, knockdown of circDDX17 could significantly enhance the migration and invasion abilities of SW480 and SW620 cells (Fig. 6f and g). The above in vitro experiments collectively indicated that silencing of circDDX17 could promote the progression of CRC cells, which was in correspondence with the clinical findings.