Clinical data
Clinicopathological data (age, sex, pathological stage, and so on) of 384 GC patients were downloaded from the The Cancer Genome Atlas 2015 RNA sequencing database (http://xena.ucsc.edu/getting-started/), together with relative expression levels of miRNAs (miR-495-3p, miR-134-5p, miR-221-3p, and miR-222-3p) in 41 paired tumor tissue samples and corresponding adjacent normal tissues. A human tissue microarray containing 90 paired tumor and corresponding adjacent normal tissues from GC patients (Cat HStmA180Sull) was purchased from the Shanghai Outdo Technology Co., Ltd. (Shanghai, China). Five frozen GC specimens from endoscopic biopsy were stored in liquid nitrogen in our laboratory. When we assessed the expression of miR-134-5p in GC, the inclusion criteria for patients were that as many GC cases as possible were downloaded from the TCGA cohort. When we assessed the association of miR-134-5p with the prognosis in GC, cases without the survival or recurrence information were excluded, and the cases left (300 GC patients) were used for clinical prognostic analysis. All experiments were approved by the Ethics Committee of Shanghai Sixth People’s Hospital (Shanghai, China). Histological type and clinical stage of specimens were classified according to the 2010 International Union against Cancer/American Joint Commission on Cancer (UICC/AJCC) TNM Staging Criteria (7th edition).
CircRNA microarray analysis
The differential expression of circRNAs was detected by circRNA microarray as previously reported [29]. In brief, total RNA was extracted from 5 paired GC tissues by using TRIzol (Invitrogen, Carlsbad, CA, USA) and quantified using a NanoDrop ND-1000 (Ultramicro spectrophotometer, NanoDrop, Waltham, MA, USA). Sample preparation and microarray hybridization were conducted according to the Arraystar’s standard protocols (Arraystar, Waltham, MA, USA).
Identification of the targets of miR-134-5p
CircNet (http://circnet.mbc.nctu.edu.tw/) was used to screen the binding sits between circPTK2 and miRNAs. We used StarBase v2.0 (http://starbase.sysu.edu.cn) to identify the targets of miR-134-5p in cancer tissues based on the rigorous screening criteria, including identification by two prediction algorithms (Pctar and miRanda), very high stringency (> 5), and expression in at least three types of cancer.
Cell culture
Normal human gastric epithelial cell line GES-1 and GC cell lines (SGC-7901, MKN-28, HGC-27, MGC-803) were obtained from the Laboratory of Digestive Disease of Shanghai Sixth People’s Hospital. All of the cells were cultured in RPMI-1640 medium (Gibco, Grand Island, NY, USA) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 100 U/mL of penicillin, and 100 μg/mL of streptomycin (Invitrogen). Cells in this medium were seeded in the culture flask and cultured in an atmosphere containing 5% CO2 and saturated humidity at 37°C.
Cell proliferation assay
Cell proliferation in GC cell lines (MKN-28, HGC-27, MGC-803, SGC-7901) was detected using a Cell Counting Kit-8 (CCK8) assay (MedChemExpress, Shanghai, China). Approximately 3 × 103 cells were seeded in each well of a 96-well plate. Each group of cells was set with 8 replicates. The edges of each well were filled with sterile phosphate-buffered saline (PBS), and cells were incubated in an atmosphere containing 5% CO2 and saturated humidity at 37°C for 0-72 h. Then, 10 μL of CCK8 solution was added directly to each well on specified date. Then, the cells were incubated at 37°C for 0.5-4 h, and the optional density was measured at 450 nm. And the cell proliferation rate was calculated by OD (OD of test group - OD of control group / OD of control group × 100%))
Colony formation assay
After GC cells were transfected with the indicated lentivirus (MKN-28 and MGC-803), mimics (MKN-28 and MGC-803), inhibitor (HGC-27 and SGC-7901), or small interfering RNA (siRNA) (HGC-27 and SGC-7901) for 48 h, 1 × 103 cells in each group were seeded in a 6-well plate and incubated at 37°C. Medium was changed every 2 days. The cell proliferation was observed, and the culture was terminated when visible colonies appeared in the 6-well plate. Then, all cells were fixed in 2 mL 4% paraformaldehyde for 15 min and stained with a crystal violet solution for 15 min. Cell colonies were then counted and analyzed.
5-Ethynyl-20-deoxyuridine (EdU) incorporation assay
After GC cells were transfected with the abovementioned plasmids for 48 h, we used a Cell-Light EdU DNA Cell Proliferation Kit (ThermoFisher, Shanghai, China) to detect the DNA synthesis according to the manufacturer’s instructions. Cells were seeded in 96-well plates (1 × 104 cells/well), with three replicates for each group. After incubation with 100 μL 50 μmol/L EdU for 2 h, 50 μL 4% paraformaldehyde was added to each well of the 96-well plates to fix the cells, and the cells were then stained with Apollo Dye Solution. PBS (100 μL) containing 0.5% TRitouX-100 was added to each well of the 96-well plates to increase the membrane permeability. Hoechst-33342 reaction solution was added to the 96-well plates to stain the nucleic acids. Olympus IX73 microscope (Olympus, Tokyo, Japan) was used to collect the images, and the proportions of EdU-positive cells were calculated.
Transwell invasion assay
After GC cells were transfected with the abovementioned plasmids for 48 h, the cell invasion assay was performed using 24-well Transwell insert chamber plates (Corning Costar, 8.0 μm pore size). The Matrigel (BD, San Jose, CA, USA) and serum-free medium were mixed on ice at a ratio of 1:4 to prepare appropriated Matrigel solution, 50 μL of which was evenly spread on the upper surfaces of transwell chamber. Cells (1 × 104/group) in 200 μL of serum-free RPMI-1640 medium were added to the upper chamber. A total of 500 μL of RPMI-1640 medium supplemented with 10% FBS was added into the lower chamber incubated at 37°C. After 48 h, the invaded cells in the upper chamber were gently wiped with a cotton swab, fixed with 4% paraformaldehyde for 15 min, stained with a crystal violet solution at room temperature for 30 min, and gently rinsed with PBS twice. The cells were then observed, photographed, counted, and analyzed.
Transfection of siRNA, overexpression lentiviruses, mimics and inhibitor
siRNAs targeting the junction region of circPTK2 (sense: 5’-AGAGGAGUGGAAGCAGAACTT-3’; antisense: 5’-GUUCUGCUUCCACUCCUCUTT-3’) and circPTK2-overexpressing lentivirus were synthesized by GenePharma (Shanghai, China). MGC-803 and SGC-7901 cells were transfected with si-circPTK2, and MKN-28 and HGC-27 cells were transfected with the circPTK2-overexpressing lentivirus using transfect-mate (GenePharma) according to the manufacturer’s instructions. miR-134-5p mimics or control mimics and inhibitor or control inhibitor were synthesized by GenePharma (Shanghai, China).
Actinomycin D and RNase R treatment
Actinomycin D and RNase R were used to detect the stability of circPTK2. In brief, 2 mg/mL actinomycin D or dimethyl suffoxide (DMSO) (Sigma-Aldrich, Sr. Louis, MO, USA) was added to the transcription of HGC-27 and MKN-28 cells, and DMSO was the negative control. After incubation for 24 h, the expression levels of PTK2 and circPTK2 were detected by quantitative real-time PCR (qRT-PCR). RNase R (3 U/μL, Epicentre Technologies, Madison, WI, USA) was added to the RNA (2 μg) of MGC-803 and MKN-28 cells respectively, and then incubated for 30 min at 37°C. Then, the RNA expression levels of PTK2 and circPTK2 were detected by qRT-PCR.
Luciferase reporter assay
Luciferase reporter assay was used to verify the binding site between circPTK2 and miR-134-5p. The luciferase reporter vectors were constructed by GenePharma. The luciferase reporter vector and internal reference were co-transfected in HEK293T cells (Chinese Academy of Sciences, Shanghai, China) in 96-well plates by using the transfect-mate transfection reagent. After 48 h, the fluorescence value was determined with a dual-luciferase reporter assay (Promega, Georgia, Madison, USA). Cell lysis (10 μL) was added to 40 μL of LAR II, then mixed, and measured the reading, which was the value of Firefly Luciferase. Then, 40 uL Stop&Glo was added and read again, which was the value of Renilla Luciferase. The ratio of two sets of data was calculated.
qRT-PCR
The RNA expression levels of circPTK2 and miR-134-5p were detected by qRT-PCR. Total RNA was extracted from the GC cells using TRIzol reagent (Invitrogen, Waltham, MA, USA). Evo M-MLV (Thermo Fisher Scientific, Waltham, MA, USA) was used to reversely transcribe RNA, and SYBR Green Master Mix kit (Takara, Otsu, Japan) was used for cDNA amplification. The amplification reaction conditions were as follows: 95°C for 30 s, 60°C for 30 s, and 72°C for 60 s. In addition, High Pure miRNA isolation kit (Roche, Basel, Switzerland) was used to isolate total miRNA, and RT-PCR was performed using a TaqMan MicroRNA Reverse Transciption kit (Life Technologies, Shanghai, China). Primers are listed in Supplementary Table S1.
Western blotting analysis
Western blotting analysis was performed to determine the protein abundance in HGC-27 and MGC-803 cells. Total protein was extracted with lysates. After centrifugation (12000 rpm, 5 min, 4℃), the supernatant fluid was collected, and the concentration was measured using a Pierce BCA kit (Thermo Fisher Scientific), followed by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Polyvinylidene difluoride (PVDF) membranes (Millipore, Boston, MA, USA) were then rinsed using a blocking solution with 5% fat-free milk for 1 h at room temperature and incubated overnight at 4°C in primary antibodies, followed by incubation in secondary antibodies for 1 h at room temperature. The anti-CUGBP Elav-like family member 2 (CELF2) (1:1000, ab186430), anti-phosphate and tension homology deleted on chromsome ten (PTEN) (1:1000, ab32199), anti-proliferating cell nuclear antigen (PCNA) (1:1000, ab92552), and anti-matrix metalloproteinase 2 (MMP2) (1:1000, ab97779, Abcam, Cambrige, England) were used as primary antibodies.
CircRNA in vivo precipitation (circRIP)
This experiment was conducted as previously reported [15]. CircPTK2-overexpressing HGC-27 cells were inoculated in a 10-cm dish. After 48 h, the specific biotin-tagged probe or control probe with a final concentration of 200 nm/L was transfected to HGC-27 cells. After 24 h, the cells were first fixed with 1% formaldehyde for 10 min, extracted with 1 mL lysate, and samples were treated with ultrasound. After centrifugation (10,000 × g, 10 min), the supernatant was transferred to a 2 mL centrifuge tube, 50 μL of it was preserved as the input control, and the remaining supernatant was used to incubate with a mixture of circPTK2-specific probe-streptavidin Dynabeads (M-280, Invitrogen) at room temperature overnight, with 200 μL of lysis buffer and proteinase K was used to reverse the formaldehyde crosslinking. Finally, the miRNAeasy Mini Kit (Qiagen, Dusseldorf, German) was used to extract total RNA according to the manufacturer’s instructions. The RNA expression levels of circPTK2 and miR-134-5p were detected by qRT-PCR.
RNA fluorescence in situ hybridization (FISH)
The expression levels and cellular localization of circPTK2 and miR-134-5p in GC tissue samples were detected by FISH analysis. A digoxin-labeled probe sequence for circPTK2 (5’- AGCCCCAGTTCTTCCACTCCTCTGG-3') and biotin-labeled probe sequence for miR-134-5p (5’-CCCCTCTGGTCAACCAGTCACA-3’) were synthesized. The FISH analysis was performed as previously reported [29].
In vivo tumor growth assay
Six-week-old NOD/SCID mice were purchased from the Shanghai Laboratory Animal Center (SLAC, Shanghai, China). MKN-28 cells (1 × 107) transfected with a circPTK2 overexpression vector or control lentivirus vector were resuspended in 200 μL of sterile PBS and injected subcutaneously into the right flanks of mice (7 mice in each group). After 30 days, the mice were euthanized, and the xenografted tumors were removed, weighed, and processed for hematoxylin-eosin (HE) staining and immunohistochemical (IHC) analysis to detect the expression of Ki-67 and CELF2. Staining intensity was accessed by using the Image-pro Plus 6.0 (Media Cybernetics) and was represented by the mean density, using the formula mena density = integrated optical density/area of interst. The animal experiments were approved by the Ethics Committee of Shanghai Sixth People’s Hospital.
Caudal vein pulmonary metastasis models
Caudal vein pulmonary metastasis models were established to detect the effect of circPTK2 on the metastasis capacity of GC cells. In brief, 1 × 107 MKN-28 cells transfected with the circPTK2 overexpression vector or control lentivirus vector were resuspended in 0.1 mL PBS and injected into the caudal vein of mice (7 mice in each group). After 30 days, the mice were euthanized, and the lung tissues were collected for HE to observe the metastasis of GC cells.
Statistical analysis
Statistical analysis was performed with GraphPad Prism 7 (La Jolla, CA, USA). Values are expressed as the mean ± standard deviation (SD). Student’s t test and analysis of variance were used for comparisons between groups. Kaplan-Meier analysis was used to assess the association of circPTK2 or miR-134-5p with the prognosis in GC. Overall survival (OS) time was calculated from the date of diagnosis. Recurrence-free survival (RFS) time is defined from the treatment to tumor recurrence. A Cox proportional hazard model was used to assess the risk of circPTK2 or miR-134-5p with respect to prognosis in GC. Only the variables with statistically significant (α < 0.05) were included in the multivariate Cox model. Pearson correlation analysis was used for the correlation between circPTK2 and miR-134-5p in GC. Categorical data were analyzed by chi-square and Fisher’s exact tests. P < 0.05 was considered statistically significant.