Bioinformatics analysis

The GSE150476 and GSE112496 datasets were used to reveal the differentially expressed circRNAs in esophageal cancer. Five circRNAs including hsa_circRNA_103225, hsa_circRNA_404013, hsa_circRNA_102471, hsa_circRNA_007624 (also termed hsa_circ_0007624 and was abbreviated as circBCAR3 in the present study), and hsa_circRNA_082546 were found to be differentially expressed in esophageal cancer based on the intersections of GSE150476 and GSE112496. The ENCORI database [32] was used to reveal the miRNAs binding to hsa_circ_0007624 and the mRNAs binding with miR-27a-3p. Under the condition of strict stringency (> = 5) of CLIP Data, miR-27a-3p was found to rank first as the target of hsa_circ_0007624. Under the condition of strict stringency (> = 5) of CLIP Data, high stringency (> = 3) of Degradome Data, more than 10 cancer types in Pan-Cancer, and more than 5 predicted programs, TNPO1, GCC2, THRB, CASC3 were found as the targets of miR-27a-3p. The binding sites of hsa_circ_0007624/TNPO1 and miR-27a-3p as well as the motif of QKI were also obtained from ENCORI. The GEPIA database [33] was used to predict the expression profile of TNPO1 and E2F7 in esophageal cancer tissues and the expression correlation between TNPO1 and QKI/E2F7. Metascape [34] was used for GO enrichment analysis of the hypoxia-induced differentially expressed RNAs in EC109 cells. Motif of E2F7 was obtained from the JASPAR database [35].

Tissue samples

Forty-five paired esophageal cancerous and adjacent noncancerous specimens (at least 5 cm away from the tumor tissues) were collected from patients by surgical operation at Ningbo Medical Center Lihuili Hospital, Ningbo University. The esophageal cancerous tissues were confirmed by histological examination. All participants had signed the written informed consents before the study. Only resected samples from patients underwent surgery with written informed consent were included. These tissue samples were immediately snap-frozen in liquid nitrogen and subsequently stored at − 80 °C. This study was approved by the Ethics Committees of Ningbo Medical Center Lihuili Hospital, Ningbo University and was performed in accordance with the principles of Declaration of Helsinki.

Cell culture

Human esophageal carcinoma cell lines EC109 (#MZ-2019, well differentiated squamous carcinoma, MINGZHOU BIO, Ningbo, China), KYSE30 (#ACC-351, well differentiated squamous carcinoma, DSMZ, Germany), KYSE70 (#ACC-379, poorly differentiated squamous carcinoma, DSMZ), KYSE150 (#ACC-375, poorly differentiated squamous carcinoma, DSMZ), KYSE180 (#ACC-379, well differentiated squamous carcinoma, DSMZ), KYSE450 (#ACC-387, well differentiated squamous carcinoma, DSMZ), and normal human esophagus epithelial cell line Het-1A (#BFN60806666, BLUEFBIO, China) were used. The cells were cultured in RPMI-1640 with 1% penicillin/streptomycin (#PM150110A, Procell, Wuhan, China) at 37 °C/5% CO2. 10% FBS (#10093, ThermoFisher) was added into the culture medium. Cells were tested for mycoplasma contamination by PCR method twice a month.

Cell transfection

CircBCAR3 overexpressing plasmid (pcDNA circBCAR3), silencing plasmid (sh-circBCAR3), and their negative controls (empty pcDNA and sh-NC) were commercially provided by GenePharma (Shanghai, China). The miR-27a-3p mimic/inhibitor and their negative controls were purchased from Ribobio Biotech (Guangzhou, China). The abovementioned plasmids and negative controls were transfected into EC109 and KYSE150 cells using Lipofectamine RNAiMax (#13778150, Life Technologies). Samples were harvested after 24 h of transfection for the further research.

Quantitative real-time polymerase chain reaction

TRIzol (#15596026, Thermo Fisher Scientific, MA, USA) was used for extraction of total RNA from cancer tissues and cells. Reverse transcription of RNAs was performed using PrimeScript RT Master Mix (Takara, Dalian, China) based on the manufacturer’s protocols. The cDNA was amplified using SYBR Premix Ex Taq (#RR420A; Takara). qRT-PCR was then conducted on an Applied Biosystems™ 7500 Fast Real-Time PCR System (#4351106). The cDNA and gDNA PCR products were observed with 2% agarose gel electrophoresis. The expression of circRNAs and mRNAs was normalized to GAPDH, while expression of miRNAs was normalized to U6. Gene expression were calculated by the 2−ΔΔCt method [36]. The primer sequences were listed as following: circBCAR3, F: 5′-CCTGGAAACAGCAATGTTGA-3′; R: 5′-GTCCATGATGTGCCTCTCCT-3′. TNPO1, F: 5′-GTCTTAACAGAGTTAGAACTTGGG-3′; R: 5′-CTTCTGGGAGTATCTTGAAAGAG-3′. QKI, F: 5′-ATTATTGGTACCTGCAGCAG-3′; R: 5′-TAGGTGCCATTCAGAATCG-3′. E2F7, F: 5′-CTCGCTATCCAAGTTATCCC-3′; R: 5′-TTTCCACACCAAGACTGAC-3′. GAPDH, F: 5′-TCAAGATCATCAGCAATGCC-3′; R: 5′-CGATACCAAAGTTGTCATGGA-3′.

Treatment of RNase R and actinomycin D

Two micrograms of total RNA was cultured with 5 U/μg RNase R (#ab286929, Abcam) for half an hour at 37 °C. EC109 and KYSE150 cells were treated with 2 μg/mL actinomycin D (#ab291108, Abcam) for 1, 2, 4, 8, 12 h. RNA expression of circBCAR3 and linear BCAR3 was analyzed using qRT-PCR.

Western blot analysis

Proteins were extracted by RIPA lysis buffer (P0013B, Beyotime), separated on 8% SDS-PAGE, transferred to a PVDF membrane (IPVH00010, Millipore, USA), and blocked with 3% skim milk powder at 37 °C for 60 min. The PVDF membrane were incubated with primary antibodies at 4 °C overnight and then incubated with appropriate HRP-labelled secondary antibodies. A ECL System (WBULS0500, Millipore) was used to develop the films. The primary antibodies against GPX4 (#52455, 1:2000) and GAPDH (#3316 s, 1:2000) were purchased from Cell Signaling Technology (Danvers, MA, USA).

Luciferase reporter assay

Dual-luciferase reporter vectors carrying the wild type (WT) fragments (5′- UAUUUUCUUAUAUACUGUGAA-3′) of TNPO1 3’untranslated region (3’UTR) or the mutant (MUT) fragments (5′-UAUUUUAGGCUAUGCACGACA-3′) were constructed and termed TNPO1-wt or TNPO1-mut. Similarly, the wild fragments (5′-TTTTCCCGCCACCT-3′) of QKI promoter that were complementary to E2F7 or the mutant fragments (5′-TCATCGTGCACCGT-3′) were subcloned into the pGL3 vector to construct the QKI-wt or QKI-mut vectors. To assess the binding between miR-27a-3p and TNPO1 3’UTR, esophageal cancer cells were co-transfected with miR-27a-3p mimics or NC mimics and the TNPO1-wt or TNPO1-mut firefly luciferase reporter vectors using a Lipofectamine 2000 kit (11,668,030, Invitrogen, USA). To assess the binding of E2F7 to QKI promoter, esophageal cancer cells were co-transfected with sh-E2F7 or sh-NC and the QKI-wt or QKI-mut vectors using a Lipofectamine 2000 kit. A renilla luciferase reporter vector was co-transfected to normalize the transfection efficiency. Luciferase activities were detected after transfection for 2 days by a Dual-Luciferase Reporter Assay System (E1980, Promega).

CCK-8, EdU, and colony formation assays

A Cell Counting Kit-8 (CCK-8, #CK04, Dojindo, Tokyo, Japan) was utilized to assess cell viability. The CCK-8 reagent (10%) was diluted to the working solution and added to a 96-well plate followed by incubation at 37 °C for 2 h. Optical density (OD) values at wavelength of 450 nm were assessed using a microplate reader (#E0226, Beyotime). The proliferation of esophageal cancer cells was measured by EdU and colony formation assays. An EdU Apollo DNA in vitro kit (#C10310, RIBOBIO, Guangzhou, China) was utilized to perform the EDU assay following the manufacturers’ instructions, and then detected under an immunofluorescence microscope. For colony formation assay, EC109 and KYSE150 cells were seeded into six-well plates with 600 cells per well. On the second week after culture, cells were fixed with paraformaldehyde (#P0099, Beyotime) and stained by 0.1% crystal violet (#C0121, Beyotime).

Wound healing assay

Monolayer esophageal cancer cells were seeded in 6 wells plate, scratched with a sterile 200 μL pipette tip, and then cultured in serum-free medium. Cells were photographed after incubation for 0 and 24 h, and the width of wounds was measured.

Transwell assays

After transfection, cells (5000 cells/well) were seeded in the inserts pre-equilibrated of 8 μm-pore Transwells (3374, Corning, USA), which were coated with (for invasion) or without Matrigel (#356237, BD Biosciences, USA) (for migration). After incubation for 24 h, cells in the upper chamber were removed by cotton swab while cells on the bottom chamber were fixed in 2% paraformaldehyde for 10 min and stained with crystal violet. Numbers of migrated or invaded cells were counted in six randomly selected fields under a microscope (ECLIPSE Ti, Nikon, Japan).

Measurement of Fe2+, MDA, lipid ROS, and GSH levels

The concentration of Fe2+ in esophageal cancer cells was measured using an iron assay kit (ab83366; Abcam) by detecting cell absorbance at 593 nm using a spectrophotometer (Thermo Fisher). The concentration of MDA was detected using a lipid peroxidation MDA assay kit (ab118970; Abcam) by detecting cell absorbance at 532 nm. To detect the level of lipid ROS, cells were stained with 10 μM C11-BODIPY581/591 probe (#D3861, Invitrogen) for 30 min. Analysis of C11-BODIPY581/591 fluorescence was conducted using a BD Accuri C6 flow cytometer (BD Biosciences). The concentration of GSH was detected using a Glutathione assay kit (CS0260; Sigma) by measuring cell absorbance at 412 nm.

Transmission electron microscopy

Esophageal cancer cells were fixed with 0.05 M cacodylate buffer (#1313, TIANDZ) containing 2.5% glutaraldehyde (#111–30-8, Merck) and 2% formaldehyde at room temperature for 1 h and then at 4 °C overnight. Next, cells were immersed in 1% osmium tetroxide, rinsed with phosphate buffer, dehydrated with gradient concentrations of ethanol, and embedded in epoxy resin. Next, samples were sliced into 70–80 nm sections using a EM UC7 ultramicrotome (Leica, Wetzlar, Germany) and stained with uranyl acetate and lead citrate. A transmission electron microscope (#H-7650, Hitachi, Tokyo, Japan) at 80 kV was utilized to observe the sections.

RNA-fluorescence in situ hybridization (FISH)

CircBCAR3 and miR-27a-3p were hybridized with Cy2-labeled probe and Cy5-labeled probe, respectively, according to the manufacturer’s protocols (GenePharma, Shanghai, China). DAPI was used for nuclear staining. A confocal laser scanning microscope (Olympus FV1000) was utilized to observe circBCAR3 and miR-27a-3p in esophageal cancer cells.

RNA immunoprecipitation (RIP) assay

A Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, USA) was used following the manufacturer’s instructions. In brief, cells were cross-linked and lysed. Lysate was treated with DNase I for 10 min and centrifuged at 12,000 g for half an hour. Sample was immunoprecipitated with QKI rabbit monoclonal antibody (ab126742, 1:1000, Abcam), or control goat anti-mouse IgG antibody (ab6708, 1:100, Abcam), and added with Protein G magnetic dynabeads (Life Technologies). After washing the beads, the immunoprecipitation was set aside for PCR analysis.

Xenograft tumor

Nude mice of both sexes (age: 6–8 weeks, weight: 22–25 g) were purchased from HUNAN SJA LABRATORY ANIMAL CO., LTD (Hunan, China). The animal study was approved by the Animal Ethic Review Committees of Ningbo Medical Center Lihuili Hospital, Ningbo University. The EC109 cells stably expressing sh-circBCAR3 or sh-nc were established by infection with corresponding lentivirus vectors (backbone: pGLVU6/Puro; #C06002; GenePharma). 1 × 106 mL− 1 (100 μL) cells were subcutaneously inoculated into the nude mice. The tumor volumes had been measured from day 5 to day 25. On day 25, the xenograft tumors were removed surgically, and the tumor weight was detected. Tumor size (width2 × length × π/6) was monitored every 5 days. Live imaging was conducted using the Xenogen in vivo imaging system, IVIS-100 (Perkin Elmer, MA, USA). All animal experiments were strictly implemented in compliance with the NIH Guide for the Care and Use of Laboratory Animals.

Lung metastasis model

BALB/c-nude mice at the age of 6 weeks were inoculated with 100 μL of single-cell suspension of EC109 cells (5 × 106/mL) via tail vein. Forty-five days later, the mice were euthanized. Lung tissues were removed for observation of lung metastasis focus.

Hematoxylin-eosin staining (H&E)

Tissues were immobilized by 4% paraformaldehyde for 24 h and embedded in paraffin. Five μm sections were collected on microslides. The sections were stained by an H&E staining kit (ab245880, Abcam) following the manufacturer’s instructions.

Immunohistochemistry (IHC)

The sections from lung tissues were dehydrated and were treated with specific primary antibodies including anti-Vimentin (ab92547; 1:200), anti-E-cadherin (ab40772; 1:500), anti-N-cadherin (ab76011; 1:200), anti-MMP2 (ab86607; 1:100), and anti-MMP9 (ab76003; 1:1000) at 4 °C overnight. Next, 100 μL of HRP conjugated second antibodies IgG (ab6721 and ab6789) at a dilution of 1:2000 were added for 60 min of incubation at 37 °C. The positive staining was visualized by a DAB kit (ab64238, abcam). The staining results were photographed with a microscopy.

Statistical analysis

Each experiment was technically repeated four times. Data were analyzed with GraphPad Prism 5.0, which was also utilized for drawing the graphs. Difference between two groups was compared using two-tailed, unpaired Student’s t-test. For multiple comparison, one-way or two-way ANOVA was performed. P values less than 0.05 indicated statistical significance.