Fig. 2

circNOX4 is upregulated in CAFs and correlates with poor prognosis of NSCLC patients. A Schematic illustration of the identification of circNOX4 upregulated in CAFs. B qRT-PCR analysis of circNOX4 expression in five paired CAFs and NFs. C qRT-PCR analysis of circNOX4 expression in CAFs, NFs, NSCLC cell lines (A549, PC9, H226, H1581), and human bronchial epithelial cells (BEAS-2B). D Diagram illustrating the genomic location and back-splicing mode of circNOX4 from the NOX4 host gene. The head-to-tail splice junction site was confirmed by Sanger sequencing using the divergent primer. E cDNA and gDNA of CAFs and NFs were amplified with convergent and divergent primers. GAPDH was the negative control. F PCR and qRT-PCR analysis of circNOX4, linear NOX4, and GAPDH in CAFs and NFs with or without RNase R treatment. G Representative images (left) and quantification (right) of circNOX4 by using RNA-FISH in NSCLC tissues with and without distant metastasis. CK19 indicated cancer cells. The nucleus (blue) was stained with DAPI. Scale bar = 40 μm, DM, distant metastasis. H Kaplan–Meier curves of OS for NSCLC patients with high (n = 48) and low (n = 36) circNOX4 expression. The cut-off value was calculated using X-tile. Statistical significance was determined by the log-rank test. I Forest plots of OS, as determined by the univariate Cox proportional hazards model. J Forest plots of OS, as determined by the multivariate Cox proportional hazards model, were constructed using variables significantly associated with OS by the univariate Cox proportional hazards model (P < 0.1). Data are expressed as mean ± SD and statistical significance was determined by two-tailed Student’s t-tests and log-rank tests. *P < 0.05, **P < 0.01 and ***P < 0.001