circFTO from M2 macrophage-derived small extracellular vesicles (sEV) enhances NSCLC malignancy by regulation miR-148a-3pPDK4 axis

Clinical samples were obtained from LC tissues and paracancerous tissues of 80 patients with NSCLC. The samples were extracted between 2016 and 2020 from patients in Department of Cardiothoracic Surgery, Xinhua Hospital, affiliated to Shanghai Jiao Tong University School of Medicine. Patients signed an informed consent before surgery. Ethics committee in Xinhua Hospital Affiliated to Shanghai Jiao Tong University supervised this investigation.

NSCLC cell lines (HCC1833, A549), normal human bronchial epithelial cells (BEAS-2B), and human monocytic leukemia cell line THP-1 (from the Shanghai Cell Bank, Chinese Academy of Sciences) used in the experiments were kept in the laboratory. We cultured the cells in RPMI-1640 medium supplemented with 10% FBS, 100 IU/mL penicillin, and 100 ug/mL streptomycin.

Experimental animals were healthy, 4–6-week-old male nude mice (BALB/c), 15 ~ 20 g. They were obtained from Shanghai Jihui Laboratory Animal Care Co., Ltd. Animal ethics committee in Xinhua Hospital approved the experiments.

After we induced the THP-1 cells into M2Ms, we washed the cells with phosphate-buffered saline (PBS) and added medium without FBS for 2 days. The culture medium of the M2Ms was harvested. We collected the EV from the culture medium by differential centrifugation as follows: medium centrifugation at 300 × g for 10 min; centrifugation of medium at 2 × g for 10 min to remove cells; centrifugation of medium at 10 × g for 0.5 h to remove cell debris; then ultracentrifugation at 120 × g for 90 min; resuspension of the EV in PBS; ultracentrifugation again at 120 × g for 90 min; finally, resuspension in PBS for collection into EP tubes. The EV’ size and concentration were quantified using a nanoparticle tracking analyzer (NTA).

Our team collected cells in logarithmic growth phase, which we inoculated into 96-well plates at 5000 cells/well. Then, 100 µL 10% CCK8 reagent was put into the corresponding wells every 12 h over the next 48 h. We incubated the cells at 37 ° away from light for 1 h. We detected OD values at wavelength λ = 450 nm by an end-point method using an enzyme labeler.

We collected cells in the logarithmic growth phase, which we inoculated into 6-well plates at 1000 cells/well. We placed them in a cell culture incubator for 2 weeks. They were subsequently fixed with 4% paraformaldehyde, stained with 0.1% Crystal Violet, washed with PBS, and analyzed using GraphPad.

We collected cells in logarithmic growth phase and inoculated them into 24-well plates at 20,000 cells/well. Cells were placed in cell culture incubator for 1 days. Then, 0.5 mL EdU solution was put to every well, and we returned plate to cell culture incubator for 2 h for EdU labeling. The EdU solution was removed, and we washed the cells with PBS. Then, we added 250 μL 4% paraformaldehyde to every well for fixation. A glycine solution was put to neutralize paraformaldehyde, and 0.5% Triton X-100 was added to promote cell permeabilization. We finally stained cells using Hoechst 33,342. We photographed the EdU-treated cells utilizing inverted fluorescence microscope and counted the EdU-positive cell numbers.

Our team collected cells in logarithmic growth phase and inoculated them into 6-well plates at 5 × 10⁶ cells/well. Cells were placed in a cell culture incubator for 1 days. Once the cells were fused to more than 95%, the original medium was discarded. The 6-well plate’s bottom was scratched perpendicularly utilizing 200-μL pipette tip. We washed non-adherent cells away with PBS. We took photographs under an inverted microscope with a 100 × field of view, and the change in the healing rate of the scratched area after 48 h was calculated.

We collected cells in the logarithmic growth phase to resuspend them in FBS-free medium at 2 × 10⁵ cells/mL. We inoculated a 200-µL cell suspension into the upper chamber and added 500 µL complete medium, including 15% FBS, to lower chamber. We cultured cells in a cell culture incubator for 1 days. Medium was discarded, and we fixed the cells in the lower chamber by adding 500 µL of 4% paraformaldehyde. The cells were then stained with 0.1% Crystalline Violet for 15 min, cleaned twice with PBS, and photographed using inverted microscope under a 100 × field of view. The stained cells were counted.

Accumulation studies have reported using human monocytic leukemia cell line THP-1 to research TAMs [22, 23]. We induced THP-1 cells into M0 and M2Ms and characterized them by RT-qPCR, western blot, and flow cytometry (Fig. 1A-B; Fig. S1A). Next, the cell culture medium (M2-CM) was collected after culturing the M2Ms in FBS-free medium for 24 h. The malignant behavior of the NSCLC cells post M2-CM and FBS-free medium treatment was examined by CCK8 assays (Fig. S1B-C), Transwell assays (Fig. S1D), and wound healing assays (Fig. S1E-F). The data showed that M2-CM significantly promoted A549 and HCC1833 cell proliferation and migration compared with the FBS-free medium group.

EV are known to be involved in IC within the TME. Therefore, we speculated whether M2-CM EV could promote NSCLC CP and metastasis. EV (M0-EV, M2-EV) was collected by ultracentrifugation from the M0-CM and M2-CM and identified (Fig. 1C-E). Next, we tested whether the NSCLC cells could take in the EV. EV were stained with PKH26 dye and added to the NSCLC cell medium for 1 days. Fluorescence microscopy confirmed that the NSCLC cells took in the PKH26-labeled EV (Fig. 1F). Transwell assays (Fig. 1G), wound healing assays (Fig. 1H-I), and CCK8 assays (Fig. 1J-K) showcased that M2-exo enhanced NSCLC cell proliferation and migration compared with the M0-exo treatment group.

NGS was used to detect circRNA expression in M0-EV and M2-EV. The data showcased that 299 circRNAs expressed differentially. Characterization showed that 142 and 157 circRNAs were up- and down-regulated, respectively (Fig. 2A). KEGG and GO pathway enrichment analyses were made utilizing the DAVID database (Fig. 2B-C). Among the enriched pathways, glycolysis-related pathways were found to promote tumor progression. Therefore, we speculated whether M2-EV could promote glycolytic progression. Western blotting confirmed that M2-EV could promote the expression of the glycolysis-related proteins HK2, LDHA, and PDK4 in NSCLC cells (Fig. 2D). Meanwhile, we found that cellular glucose uptake, ATP levels, and lactate production increased after treatment with M2-EV (Fig. 2E). Next, the Seahorse assay was employed to verify the M2-EV effects upon glycolysis in the NSCLC cells. The outputs showed that the cellular respiration level decreased and the glycolysis level increased after treatment with M2-EV (Fig. 2F-G and Fig. S2A-B). PET-CT scanning indicated that the M2-EV treatment could significantly enhance glucose metabolism in subcutaneous tumors in mice (Fig. 2H). The above data indicated that M2-EV promoted the glycolytic process in NSCLC cells.

Five upregulated circRNAs in NSCLC cells with or without M2-EV treatment were selected and verified by RT-qPCR (Fig. 3A). The outputs showed that has_circ_0005941 expression was significantly increased. Hsa_circ_0005941 was generated by the FTO gene located on chr16:53,907,697–53922863. The circFTO back-splice site was verified by Sanger sequencing (Fig. 3B). Divergent primers for circFTO could be amplified via PCR from cDNA yet not gDNA (Fig. 3C). Next, we examined circFTO expression in NSCLC cell lines to find that it expressed highly in NSCLC cells. The highest expression was in A549 and HCC1833 cells (Fig. 3D). Therefore, both A549 and HCC1833 cells were selected for the following studies. Compared with linear FTO mRNA, circFTO was more stable post RNase R and actinomycin D treatments (Fig. 3E-F).

CircFTO expression in NSCLC tissues was verified through FISH. Outputs showcased that circFTO expression incremented in LC tissues (80 cases) comparing to the adjacent normal tissues (Fig. 3G). circFTO expression level also correlated to the tumor’s pathological grade and TNM stage (Table S2). Kaplan–Meier analysis of correlations between circFTO expression and overall survival revealed that patients having a high circFTO expression had a poor prognosis (Fig. 3H).

To confirm whether M2-exo regulates NSCLC progression via circFTO, we established a stable circFTO-knockdown THP-1 cell line by lentiviral transfection. Transfection efficiency was verified through RT-qPCR (Fig. S3A). The circFTO-knockdown THP-1 cells were induced to the M2 type, and their secreted EV were collected. Sh-circFTO-M2-exo and M2-exo were added to A549 and HCC1833 cell media, respectively. CCK8 assays and Transwell assays (Fig. S3B-C) showed that downregulating circFTO decreased the promotion effect of circFTO on NSCLC cell malignant progressions, including migration and proliferation. The above results demonstrated that M2-exo enhanced NSCLC cell proliferation and migration via the circFTO delivery.

Our team established stable circFTO-knockdown cell lines by lentiviral transfection and verified the effectiveness of circFTO-knockdown by RT-qPCR (Fig. S4A). Wound healing and Transwell assays showed that circFTO-knockdown significantly suppressed NSCLC cell migration ability (Fig. 4A-C). Meanwhile, CP was assessed by clone formation assays (Fig. 4D), EdU assays (Fig. 4E-F), and CCK8 assays (Fig. 4G-H). The data showed that circFTO-knockdown inhibited CP significantly. The findings unraveled that knocking down circFTO expression inhibited NSCLC cell proliferation and migration.

To evaluate circFTO effects on NSCLC cells in vivo, we injected circFTO-knockdown HCC1833 cells and normal HCC1833 cells subcutaneously into BALB/cA nude mice. In this subcutaneous tumor formation model, circFTO-knockdown suppressed tumor growth in volume and weight compared to control group (Fig. 5A-C). Simultaneous IHC staining showcased that Ki-67 expression level within tumors was decreased in circFTO-knockdown group (Fig. 5D). In a caudal vein transfer model, bioluminescence imaging showed a significantly lower fluorescence intensity and ratio in lungs of circFTO-knockdown group than control group (Fig. 5E). Meanwhile, HE staining demonstrated that circFTO knockdown significantly reduced the metastatic nodule numbers in lungs (Fig. 5F). In studies of the plantar lymph node metastasis model, bioluminescence imaging showed that circFTO knockdown significantly inhibited lymph node metastasis of the plantar tumors (Fig. 5G), and IHC staining showed that CD31 expression decreased (Fig. 5H). The above results suggested that circFTO knockdown inhibited NSCLC proliferation and metastasis.

We hypothesized that circFTO might function as miRNA sponge in NSCLC. miRNA sequencing showed that circFTO-knockdown resulted in different expressions of miRNA in HCC1833 cells (Fig. 6A). KEGG pathway enrichment analyses similarly showcased that metabolic pathways were affected significantly (Fig. S4B). To identify the downstream miRNAs of circFTO, RNA-seq was performed on the circFTO-knockdown HCC1833 cells and cross-analyzed with the predicted results from two miRNA target prediction databases (circBANK, ENCORI). The output data showcased that miR-148a-3p was the only circFTO downstream target (Fig. 6B). Targeting correlation between miR-148a-3p and circFTO was verified through LR assays (Fig. 6C-D). The outputs showed that miR-148a-3p mimics significantly inhibited circFTO-WT LA, though LA of circFTO-MUT did not alter significantly. The ENCORI database showed that patients having higher miR-148a-3p expression levels have longer survival times (Fig. 6E). To validate the miR-148a-3p effects upon circFTO expression, HCC1833 and A549 cells were transfected with mimics and miR-148a-3p inhibitor, respectively. Transfection efficiency was verified by RT-qPCR (Fig. 6F). We performed salvage assays to further research miR-148a-3p and circFTO functions during NSCLC progression. CCK8 assay results showed that the inhibitory effect caused by circFTO knockdown could be reversed post miR-148a-3p inhibitor transfections (Fig. 6G). Similarly, the Transwell assays illustrated that miR-148a-3p inhibitors could reverse NCSLC cell migration inhibition ability via circFTO knockdown (Fig. 6H). The data revealed that miR-148a-3p functioned critically in circFTO-mediated regulation effect on the progress of NSCLC.

To investigate molecular mechanism of miR-148a-3p in NSCLC, miR-148a-3p downstream targets were analyzed utilizing RNA-seq combined with bioinformatics analysis results from Targetscan, ENCORI, and miRwalk, which showed 28 possible genes targeted through miR-148a-3p (Fig. 7A-B). The glycolysis-related protein PDK4 was found among the 28 proteins, so we selected PDK4 for downstream analysis. Next, mutant and wild-type luciferase report analysis plasmids for the PDK4 3'-UTR were made. Data showcased that miR-148a-3p mimic transfections significantly suppressed PDK4-WT (wild) fluorescence intensity, whereas there was no significant alteration in PDK4-MUT (mutant) luciferase activities (Fig. 7C-D). RT-qPCR and western blot data demonstrated that PDK4 was regulated negatively by miR-148a-3p (Fig. 7E-F). In addition, circFTO-knockdown (shcircFTO) similarly inhibited mRNA and protein levels with regard to PDK4, and miR-148a-3p inhibitor ultimately rescued this inhibition (Fig. 7G-H). In conclusion, our results confirmed that circFTO upregulated PDK4 expression through the sponge miR-148a-3p.

PDK4 positively regulates glycolysis in some tissues and cancers. To research whether circFTO could promote NSCLC cell glycolysis and progression through miR-148a-3p/PDK4 axis, we performed a Seahorse assay and relevant functional tests relating to glucose metabolism. The result found that circFTO-knockdown suppressed glucose uptake, ATP levels, and lactate production in NSCLC cells. Also, a miR-148a-3p suppressor was able to reverse this process, while miR-148a-3p mimic inhibited glucose uptake, lactate, and ATP production in NSCLC cells (Fig. 8A-C). Meanwhile, Seahorse assay data showed that circFTO knockdown or miR-148a-3p overexpression increased cellular respiration levels and decreased glycolysis levels. The miR-148a-3p inhibitor could reverse circFTO-knockdown effects completely (Fig. 8D-E). These results demonstrate that circFTO could promote NSCLC cell glycolysis and progression through the miR-148a-3p/PDK4 axis.

Similarly, our team found that glycolysis, ATP levels, and lactate production in NSCLC were attenuated by sh-circFTO-M2-EV compared to M2-EV treatment (Fig. S5A-C). The Seahorse assay also showed the same results (Fig. S5D-E). The above results suggest that M2-exo promoted NSCLC glycolysis and progression via the circ-FTO/miR-148a-3p/PDK4 axis.