2.1 Clinical samples

After surgical resection, NSCLC tissue samples (n = 40) and normal tissue samples (n = 40) were collected from Kaihua County Traditional Chinese Medicine Hospital. The distance of normal lung tissue samples from the margin of cancer tissue was ≥ 2 cm, and the patient did not undergo any preoperative chemoradiotherapy. The clinical characteristics of the patients are shown in Table 1. According to the Response Evaluation Criteria in Solid Tumors criteria, PTX-sensitive tumors were defined as those in which a ≥ 30% reduction in the largest lesion was observed, whereas PTX-resistant tumors were those that did not show any significant change after PTX treatment. All specimens were frozen in liquid nitrogen and kept at −80 ℃. This study obtained patients’ informed consent and was approved by the Research Ethics Committee of Kaihua County Traditional Chinese Medicine Hospital (No. 2019Z0611).

Table 1 Correlation between circPDK1 level and pathological indexes of NSCLC patients2.2 Cell lines

Primary human bronchial epithelial cells (HBE) and NSCLC cell lines (H1299, A549, H-125 and NCI-H292) were provided by Procell (Wuhan, China). HBE cells were cultured in DMEM (Procell) containing 10% FBS (Procell) and 1% penicillin/streptomycin (Sigma-Aldrich) with 5% CO2 and 95% air. H1299 and A549 cells were put in RPMI-1640 medium (GIBCOBRL, USA) + 2 mM glutamine, and H-125 and NCI-H292 in RPMI-1640 medium (Invitrogen, USA) + 10% FBS (Invitrogen) + 1% penicillin/streptomycin (Sigma-Aldrich). PTX (Sigma) was dissolved in DMSO (Sigma). The PTX-resistant H1299/PTX cell line was established by exposing H1299 cells to increasing concentrations of PTX (0.1–0.5 μM) until the ability of the cells to grow in the presence of PTX at the same rate as parental cells in the absence of the drug. H1299/PTX cells were cultured in DMEM/F12 medium supplemented with 10% FBS (containing 5 μM PTX, 100 U/mL penicillin, and 100 mg/mL streptomycin) and routinely incubated overnight (5% CO2, 37 °C) [30].

PTX-resistant A549/PTX cell line was established by exposing A549 cells to increasing concentrations of PTX (50 nM-250 nM). To maintain the PTX-resistant phenotype of A549/PTX cells, 0.1 μM PTX was added to the culture medium and routinely cultured overnight (5% CO2, 37 °C) [31].

2.3 Cell transfection

Cell transfection was achieved in H1299/PTX cells and A549/PTX cells by Lipofectamine 2000 (Invitrogen). si-circPDK1, si-NC, miR-4731-5p mimics/inhibitor, miR-NC, pcDNA3.1-GIGYF1, and pcDNA3.1-vector were prepared by GenePharma (Shanghai, China). RT-qPCR was conducted to evaluate the transfection efficiency.

2.4 RT-qPCR

Based on Trizol reagent (Thermo, USA), total RNA was extracted from NSCLC tissues and cells, of which the concentration was determined using BCA kits (Invitrogen). Then, miRNA cDNA was generated with miRNA reverse transcription kit (TaKaRa), and mRNA/circRNA cDNA with PrimeScript™RT Reagent kit (TaKaRa). Next, three repeats of RT-qPCR were completed for each sample on the CFX96 Contact real-time fluorescent quantitative PCR assay system (Bio-Rad, USA) using the SYBR Green PCR Premix kit (Invitrogen). RNA expressions were calculated using 2−ΔΔCt and normalized to GAPDH or U6. The primers are shown in Table 2.

2.5 CCK-8 analysis

H1299/PTX cells and A549/PTX cells were placed into 96-well plates at 2 × 103 cells/well. After the specified time (0, 24, 48, and 72 h), 10 μL CCK-8 (Beyotime, Shanghai, China) was supplemented and reacted for 1 h before reading optical density at 450 nm on a microplate reader (Bio-Rad).

H1299/PTX cells and A549/PTX cells at logarithmic growth were placed in 96-well plates (1 × 104 cells/well) after digestion and cultured for 24 h. Then, 0.05, 0.2, 0.8, 3.2, 12.8 μM PTX was added and reacted for 48 h. Afterward, cells were combined with 10 μL CCK-8 reagent at 37 ℃ for 2 h, and absorbance (450 nm) was read on a microplate reader (Bio-Rad). The growth curve was plotted using Graphpad Prism to calculate the IC50 value of PTX.

2.6 Colony formation test

H1299/PTX cells and A549/PTX cells were plated in 6-well plates in an amount of 5 × 103 cells per well and cultured for 36 h in 10% FBS-DMEM (37 ℃, 5% CO2). Then fresh medium + 2 μM PTX were replaced and cells were stimulated for 7 days. Cell culture was terminated upon the appearance of colonies. Then, cells were rinsed twice with PBS (Beyotime), fixed with 4% formaldehyde (Beyotime) for 10 min, stained with 5% crystal violet (Sigma) for 10 min, and calculated by microscopy (Olympus, Japan).

2.7 Flow cytometry

Apoptosis rate was assessed by Annexin V-FITC/PI Apoptosis Detection kit (Invitrogen). H1299/PTX cells and A549/PTX cells (1 × 105 cells) were suspended in a 1 × Binding Buffer (500 μl), further reacted with Annexin V-FITC (5 μl) and PI (5 μl) for 15 min without light, and detected on a FACScan® flow cytometer (BD Biosciences, USA).

2.8 Transwell experiment

The upper chamber of Transwell chamber (Costar, USA) was coated with Matrigel, in which 200 μL serum-free DMEM at cell concentration of 5 × 106 cells/ml was added. Meanwhile, 600 μL 10% FBS-DMEM (HyClone) was added to the lower compartment and cultured for 24 h (37 ℃, 5%CO2). Next, non-invasive cells were removed, and the remaining cells after PBS treatment were fixed with 4% paraformaldehyde (Beyotime), treated with 0.5% crystal violet (Sinopharma), and observed by inverted microscopy (Olympus) in 5 fields.

2.9 Dual luciferase reporter experiment

Circbase (http://circbase.org/) and TargetScan (www.targetscan.org) were utilized to predict the potential binding sites of miR-4731-5p, circPDK1 and GIGYF1. GenePharma synthesized wild-type or mutant circPDK1 fragments and GIGYF1 containing miR-4731-5p binding sites. The fragments were inserted into pmirGLO vector (Promega) to complete the construction of circPDK1-WT, GIGYF1-WT, circPDK1-MUT, and GIGYF1-MUT. H1299/PTX cells and A549/PTX cells were put in 96-well plates with 1 × 104 cells/well and transfected with luciferase reporter and miR-4731-5p mimic or mimic-NC using Lipofectamine®3000 (Invitrogen) for 48 h. Assays of luciferase activity were conducted using a dual luciferase assay system (Promega).

2.10 RIP experiment

RIP was determined using Magna RIP kit (Millipore). A lysate was produced by lysing H1299 cells and A549 cells with RIP lysis buffer and incubated with magnetic beads containing anti-AgO2 (Abcam, USA) or anti-IgG (Abcam) at 4 ℃ for 6 h. The immunoprecipitate bound to magnetic beads was eluted with elution buffer, and circPDK1, miR-4731-5p and GIGYF1 were analyzed by RT-qPCR after purification of RNA.

2.11 Xenotransplantation of nude mice

Animal procedures were approved by the Kaihua County Traditional Chinese Medicine Hospital Animal Care and Use Committee (No. 201909ZJ41). Twenty 6-week-old BALB/c nude mice (Shanghai SLAC Laboratory Animals Co., Ltd., Shanghai, China) were randomly divided into si-NC group, si-NC + PTX group, si-circPDK1 group, and si-circPDK1 + PTX group. H1299 cells stably transfected with si-circPDK1 and si-NC were re-suspended in PBS buffer (1 × 105 cells/ml), and nude mice were given 200 mL of H1299 cell suspension subcutaneously in the left armpit. A week later, 3 mg/kg PTX or PBS injections were administered intraperitoneally to mice every 3 days. In a pathogen-free environment at 22 ± 3 °C with 50 ± 15% relative humidity, 12 h of light/dark cycle were maintained for the mice. Every 7 days, tumor volume was measured and calculated using a caliper (short diameter2 × long diameter × 0.5). 35 days later, tumors were resected from euthanized mice.

2.12 Immunohistochemistry

Tumors were fixed overnight in 10% formalin (Thermo Fisher Technology), embedded in paraffin, and sliced to 4 μm. After dewaxing with xylene (Beyotime), tissue sections were blocked by 0.3% H2O2 for 10 min and bound with PBS containing 0.3% Triton X-100 and 5% FBS for 1 h. Next, GIGYF1 antibody (ab121784, Abcam) was reacted at 4 °C overnight, as well as the secondary antibody (1:1000; ab6720, Abcam) for 1 h. Sections were developed with DAB (Vector Labs, USA) and re-stained with hematoxylin for 2 min, followed by microscopic imaging (Leica).

2.13 Western blot

Tissues and cells were lysed on ice with RIPA lysis buffer (Vazyme, FD008) for 20 min, and protein concentrations were detected with Pierce BCA Protein Assay Kit (Rockford). The protein was isolated using 10% SDS-PAGE, transferred to a PVDF membrane (Millipore), sealed with 5% skim milk for 2 h, and treated with GIGYF1 (Abcam; ab121784) and GAPDH (Abcam; ab37168), Ki-67 (Abcam; ab270650), E-cadherin (Abcam; ab233611), N-cadherin (Abcam; ab254512) overnight at 4 °C, as well as secondary antibody (Abcam; ab205719) at 37 ℃ for 1 h. Results were developed with enhanced chemiluminescence detection kit (Vazyme; E411-04) and analyzed in the FluorChem™ system.

2.14 Data analysis

Experimental data underwent statistical analysis based on SPSS20 statistical software. Measurement data were presented as ˉx ± s and tested by t-test or one-way ANOVA. P < 0.05 or < 0.01 indicated statistically significant differences.