Samples collection

The open-accessed circRNAs expression profile and corresponding tissue information were obtained from the Gene Expression Omnibus (GEO), GSE112214 and GSE158695 projects. Detailed, the GSE112214 and GSE158695 projects both provide the circRNAs expression profile of three NSCLC and normal adjacent tissues (NATs), whose annotation platform is GPL19978. Eighty patients with NSCLC and normal adjacent tissues (NATs) underwent surgical resection at the Huadong hospital. Totally, 39 NSCLC samples and 20 NATs were collected fromthe Huadong hospital. The samples were promptly preserved in liquid nitrogen at the Huadong hospital. Each sample underwent independent evaluation by two distinct pathologists. Importantly, no patients had been subjected to any preoperative procedures. We also gathered preoperative serum samples from each participant. The Ethical Committee of Huadong hospital, approved the use of these specimens. Written informed consent was procured from every participating patient. Pan-cancer data was get from the UCSC Xena database (https://xenabrowser.net/datapages/). Expression profile and clinical data of NSCLC patients in The Cancer Genome Atlas Program (TCGA) database was downloaded from the TCGA-GDC [16]. Open-accessed clip-seq data was get from the ENCORI database (http://starbase.sysu.edu.cn/).

Cell lines

The human non-small cell lung cancer (NSCLC) cell lines H522, A549, H1299, H460, the human bronchial epithelium cell line (BEAS-2B),human monocytes cell line THP-1 and mouse Lewis lung carcinoma cell line (LLC), were all obtained from the American Type Culture Collection (ATCC, Manassas, VA). These cell lines were all grown in a controlled environment at 37 °C with a 5% CO2 concentration, using RPMI 1640 medium (Biosharp, Guangzhou, China) or DMEM medium (Biosharp, Guangzhou, China). The media were further supplemented with 10% fetal bovine serum (FBS; Gibco, South America).

Animal experiments

Every animal experiment conducted received approval from the Fudan University (20230304Z). Both subcutaneous xenograft and tail vein lung metastasis models were established using female BALB/c nude mice. For the subcutaneous tumor model, roughly 2 × 106 A549 cells suspended in a 40% Matrigel (BD, San Jose, CA, USA) medium were injected into each mouse's flank. Tumor development was monitored and quantified on a weekly basis, with tumor volume calculated using the formula (length × width2/2). After a period of four weeks, the mice were euthanized and tumor weights were recorded. For the tail vein metastasis model, an approximate of 1 × 106 A549 cells were introduced via the tail vein into the nude mice. Four weeks later, all mice were euthanized. Their lungs were imaged and then harvested for ensuing analysis. LLC cells were utilized to establish a tail vein metastasis model in C57/B6 mice. For the exosome treatment experiments, exosomes produced by both control and A549 cells, which overexpressed circATP9A, were administered intravenously into the tail vein of C57/B6 mice. Each injection contained 2 μg of exosomes, and the mice received two injections per week. After a total of five injections, the mice were sacrificed for tissue collection and further assessments. Ki67 expression in tumor tissues from xenografted nude mice was quantitatively assessed using the H score. Tumor samples were first fixed, paraffin-embedded, and sectioned for immunohistochemical (IHC) analysis. These sections were stained with a Ki67-specific antibody, highlighting proliferating cells. The H score for Ki67 was calculated by evaluating both the intensity of staining (graded as 0, 1 + , 2 + , or 3 + for none, weak, moderate, or strong staining, respectively) and the percentage of Ki67-positive cells in the tissue. The final H score, a product of the intensity score and the percentage of positive staining.

Plasmid construction and cell transfection

For stable circATP9A knockdown, three unique shRNAs targeting circATP9A were engineered into lentiviral vectors, a task completed by GenePharma (Suzhou, China). A nonspecific sh-NC vector was employed as the control. Additionally, a full-length circATP9A lentiviral vector, also synthesized by GenePharma, was employed for circATP9A overexpression, with a control vector lacking any circATP9A sequence used as the reference group. For the generation of stable cell lines, A549 and H1299 cells were infected with respective lentiviruses using 1 µl of Polybrene (5 µg/µl) from GenePharma as an enhancer. After 72 h, the cells were exposed to 5 µg/ml of puromycin in medium for selection. Post a 10-day duration, puromycin-resistant cells were deemed as stably transfected. A similar methodology was applied for siRNAs against HuR and NUCKS1-OE plasmids, both sourced from RiboBio (Guangzhou, China). The transfection of si-RNAs and plasmids was executed using Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA), abiding by the manufacturer’s instructions. The knockdown or overexpression efficacy of specific molecules was validated through qRT–PCR or western blot assays. The sequences of sh-circATP9As and siRNAs are provided in Table S1.

Quantitative Real-time PCR (qRT–PCR)

Total RNA was extracted using the TRIzol reagent (Invitrogen, Carlsbad, CA, USA), followed by cDNA conversion via the PrimeScript RT Reagent Kit (Takara, Tokyo, Japan). The ensuing qRT-PCR analysis was carried out utilizing the Sybr green system, with GAPDH functioning as the internal standard. RNA relative expression levels were computed via the 2–ΔΔCT method. Table S2 lists the primer sequences used in this study.

Rnase R treatment

In summary, RNA was retrieved from A549 and H1299 cells with TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the provided instructions. This was accompanied by the addition of Rnase R (1.0 U/μg) (Geenseed, Guangzhou, China) to 500-ng RNA, and incubation for 20 min at 37 °C. For control, a similar amount of RNA was processed without Rnase R under identical conditions. The stability of circATP9A versus linear mRNA-ATP9A was then assessed using qRT-PCR.

Electrophoresis analysis

For gel electrophoresis, a 1% agarose gel was prepared by heating 50 ml of 1 × TAE buffer with 0.5-g agarose until boiling. After cooling to 70 °C–80 °C, 5 μl of 4S GelRed (Sangon Biotech, Shanghai, China) was thoroughly mixed in. After solidification of the poured solution in a mold, the gel was placed in an electrophoresis tank filled with 1 × TAE buffer. Subsequently, 10 μl of DNA samples, mixed with loading buffer, were loaded into each well. The electrophoresis ran at 120 V for 30 min, and the resulting bands on the gel were captured using an ultraviolet imaging system.

In situ hybridization (ISH)

For ISH, we used a probe targeting the circATP9A splicing site, tagged with 5′-digoxin (DIG) and 3′-DIG. This probe was custom made by Servicebio (Wuhan, China). Scramble probe (negative) and a U6 probe (internal) served as controls. The procedure started with deparaffinization and rehydration of paraffin sections with xylene and a series of graded ethanol. Subsequently, a 20-min incubation with proteinase K at 37 °C was performed, followed by a 10-min Triton-X100 treatment at 4 °C. The sections were then incubated overnight at 37 °C in a hybridization buffer containing the circATP9A probe, and another overnight incubation with the anti-digoxin antibody at 4 °C followed. Afterwards, sections were treated with 5-Bromo-4-Chloro-3-Indolylphosphate/Nitroblue Tetrazolium (BCIP0/NBT) (Beyotime, Shanghai, China) at room temperature for 30 min, and with Nuclear fast red (Servicebio, Wuhan, China) at room temperature for 3 min. Visualization and imaging were done with an Olympus microscope (Olympus, Tokyo, Japan). The H-score for circATP9A was computed as: H-score = Σ (P × I), where P signifies the percentage of stained cells, and I represents the staining intensity score: 0 for no staining, 1 for weak, 2 for moderate, and 3 for intense staining. Table S3 provides details of the ISH probes.

RNA fluorescence in situ hybridization (FISH)

To ascertain the subcellular localization of circATP9A in NSCLC cells, we used a Cy3-labeled circATP9A-specific probe from RiboBio (Guangzhou, China). Briefly, NSCLC cells (A549 and H1299) were trypsinized and resuspended in medium. Approximately 2000 cells were seeded on a 48-well plate with a glass cover slip. Upon reaching 70% to 90% confluence, cells were thrice washed with PBS and fixed with 3.7% paraformaldehyde. Cell permeabilization was achieved using 0.5% Triton-100 for 10 min at 4 °C. Cells were then pre-hybridized at 37 °C for 30 min, followed by overnight incubation at 37 °C with the circATP9A-FISH probe in a dark hybridization buffer. After hybridization, cells were washed using SSC solutions and stained with DAPI for 15 min in darkness. The prepared slides were observed and photographed using a confocal fluorescence microscope (Carl Zeiss AG, Jenna, Germany). Table S3 contains information about the circATP9A-FISH probe utilized in this experiment.

Colocalization of circATP9A with HuR

Fluorescence staining was employed to assess the colocalization of circATP9A and HuR in NSCLC cells. Briefly, around 2,000 NSCLC cells were seeded onto a 48-well plate with cover glass. Upon reaching 70%–90% confluence, the cells were washed three times with PBS, then fixed with 3.7% paraformaldehyde. Subsequently, cells were permeabilized using 0.5% Triton-100 for 10 min at 4 °C. Pre-hybridization for 30 min at 37 °C using pre-hybridization buffer was carried out, followed by an overnight incubation at 37 °C with a Cy3-labeled circATP9A-FISH probe (RiboBio) in hybridization buffer in a dark environment. The cells were then permeabilized again with 0.5% Triton-100 for 10 min at 4 °C, before incubating with an anti-HuR antibody (Abcam) overnight at 4 °C in darkness with gentle rotation. Following this, cells were washed with PBS and nuclei were stained with DAPI for 15 min. Finally, cells were visualized and imaged with a confocal fluorescence microscope (Carl Zeiss AG, Jenna, Germany).

Western blotting

The Total Protein Extraction Kit (KeyGEN, Nanjing, China) was used to extract the total protein in NSCLC cells, and the concentrations were measured by a BCA protein assay kit (KeyGEN, Nanjing, China). Antibodies against HuR (ab200342, 1:1000), NUCKS1 (12,023–2-AP, 1:1000), β-actin (20,536–1-AP, 1:5000), p-AKT (Ser273) (66,444–1-Ig, 1:5000), AKT (60,203–2-Ig, 1:5000), p-mTOR (Ser2448) (67,778–1-Ig, 1:2000), mTOR (66,888–1-Ig, 1:5000), CD9 (20,597–1-AP, 1:1000), CD63 (25,682–1-AP, 1:1000), CD81 (66,866–1-Ig,1:2000), Calnexin (10,427–2-AP, 1:5000), hnRNPA2B1 (14,813–1-AP, 1:2000) were obtained from Abcam and Proteintech. Chemiluminescent signals were detected using Western ECL Substrate (Advansta, Menlo Park, CA, USA) and images were captured with a ChemiDoc Imaging System (Bio-Rad, Hercules, CA, USA).

RNA pull-down assay

The interactions between circATP9A and HuR, as well as circATP9A and hnRNPA2B1, were verified using biotin-coupled circATP9A probes and control probes, furnished by GenePharma (Suzhou, China). The protocol is outlined as follows: Roughly 1 × 107 A549 and H1299 cells were lysed and sonicated in a 4 °C water bath for half an hour. A fraction (20 µl) of the lysate was allocated for RNA input, while a major portion (80 µl) was set aside for protein input. Probes were then incorporated into the lysate and stirred at room temperature for 16–24 h. Next, 100 µl of streptavidin magnetic beads (MCE, Monmouth Junction, NJ, USA) were added to the lysate and rotated at room temperature for 2–4 h. Using a magnetic stand, the beads were gathered and washed five times with washing buffer (containing PMSF, Protease inhibitor, and Rnase inhibitor). The cleansed beads were resuspended in 1 ml of washing buffer, with 100 µl of this solution used for RNA purification and the remaining 900 µl allocated for protein purification. For RNA extraction, the 100-µl sample was combined with 5 µl of proteinase K (Sangon Biotech, Shanghai, China) and RNA PK buffer, then gently rotated at 50 °C for 45 min followed by 10-min heating at 95 °C to break the formaldehyde cross-links. The RNA was subsequently purified using TRIzol reagent (Invitrogen, Carlsbad, CA, USA), converted into cDNA, and stored at -80 °C for future use. For protein extraction, 300 µl of 4 × loading buffer was added to the remaining 900-µl sample, followed by a 10-min incubation at 100 °C. The supernatant with the protein extract was then isolated using the magnetic stand, and the protein was deployed for mass spectrometry (MS) analysis and Western blotting. The sequences of the circATP9A probe and the control probe are presented in Table S3.

Silver staining

Overall, proteins extracted from the circATP9A RNA pull-down assay were differentiated using a 10% SDS-PAGE gel. The gel was subsequently stained with a Silver Stain kit (BL620A, Biosharp, Beijing, China) in accordance with the producer’s instructions.

RNA immunoprecipitation (RIP) assay

For verification of the link between HuR and circATP9A, and hnRNPA2B1 and circATP9A, a RIP assay was conducted employing a RIP kit (Millipore, MA, USA). In essence, approximately 2 × 10^7 A549 and H1299 cells were harvested and lysed with RIP lysis buffer. Interacting RNAs were then precipitated with anti-HuR antibody (ab200342, Abcam) and anti-hnRNPA2B1 antibody (14,813–1-AP, Proteintech). The anti-IgG antibody (ab172730, Abcam) was utilized as a negative control. The co-precipitated RNAs were then isolated using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and their abundance was evaluated by qRT-PCR.

Co‐culturing system

To mimic the EV-mediated intercellular communication that takes place between tumor cells and TAMs, an in vitro indirect co-culture model was established. Macrophages and LUAD cells were individually planted into the upper and lower chambers of a Corning® Transwell® cell culture insert (4 μm pore size, Corning Inc., Corning, NY, USA), incorporating a polycarbonate membrane. After a 48-h co-culture period, the cells were harvested for further experiments.

Isolation Evs from cell medium

Evs were derived from NSCLC cell media through a series of steps: NSCLC cells were cultured in a medium enriched with 10% EV-free FBS. Following a 72-h incubation at 37 °C with 5% CO2, the medium was collected and centrifuged under various conditions: initially at 2,000 g for 10 min, then at 3,500 g for 20 min, followed by 10,000 g for 1 h, and finally at 120,000 g for 2 h. All centrifugation phases were carried out at 4 °C. The resultant purified Evs were re-suspended in PBS and then stored at -80 °C for subsequent use.

Macrophage induction from monocytes and flow cytometry

THP1 cells, cultured in six-well plates, were treated with 100 ng/mL of phorbol-12-myristate-13-acetate (PMA; Sigma-Aldrich) and incubated for 24 to 48 h. After incubation, the medium was swapped with fresh PMA-free medium, and the cells were maintained for another 3 days before use. For the detection of macrophage surface markers, cells in chilled PBS were treated with either anti‐CD206 or anti‐HLA-DR antibodies (both from eBioscience) at 4℃ for 30 min. Following incubation, the cells were washed and subsequently analyzed using a BD Accuri™ C6 flow cytometer (BD Biosciences, San Jose, CA, USA) to detect macrophage surface markers CD206 and HLA-DR.

Evs internalization

Evs derived from RCC cells were labeled using the PKH26 Red Fluorescent Cell Linker Kit (Umibio, Shanghai, China). Following this, PMA-stimulated THP-1 cells [THP-1 (Mφ)] were co-incubated with these PKH26-tagged Evs in a dark setting overnight. After staining the cells with 4′,6-diamidino-2-phenylindole (DAPI), the EV uptake process was visualized using a confocal fluorescence microscope (Carl Zeiss AG, Jenna, Germany).

CCK-8 assay, EdU assay, colony formation assay, and transwell assay

The proliferation of NSCLC cells was gauged using the CCK-8 assay, the 5-ethynyl2′deoxyuridine (EdU) assay, and the colony formation assay. The cell invasion and migration capabilities were assessed through transwell invasion and migration assays, executed following the protocols delineated in our previous research [17].

Statistics analysis

Statistical evaluations were conducted using GraphPad Prism and SPSS Software. A two-tailed Student’s t-test, ANOVA followed by Tukey’s multiple comparisons post-test, and Pearson’s correlation analysis were the methods used for statistical comparisons. All statistical data are expressed as mean ± standard error of the mean. All p values were calculated using a two-sided test, with p values < 0.05 considered statistically significant. Each experimental procedure was performed at least three times for reliability. The survival difference in different groups was compared using the Kaplan–Meier method.