Circ_0074158 and QKI6: A regulatory axis for PD-L1 ubiquitination and immune evasion in NSCLC

Abstract

Objective

CircRNAs are involved in cancer progression. However, their role in immune escape in non-small cell lung cancer (NSCLC) remains poorly understood.

Methods

This study employed RIP-seq for the targeted enrichment of circRNAs, followed by Western blotting and RT-qPCR to confirm their expression. Specific binding was assessed using electrophoretic mobility shift assays. Transwell and wound healing assays were performed to evaluate the invasive and migratory capacities of NSCLC cells. The secretion of TNF-α, IFN-γ, GzmB, and perforin by CD8⁺ T cells co-cultured with peripheral blood mononuclear cells (PBMCs) was quantitatively measured using enzyme-linked immunosorbent assays. The PD-L1 phenotype and apoptosis levels were determined by flow cytometry, while cell proliferation and apoptosis were evaluated through CCK-8, EdU, and TUNEL assays. The in vivo role of circ_0074158 was investigated using a mouse subcutaneous tumour implantation model.

Results

QKI6 modulates the level of PD-L1 ubiquitination in NSCLC. Both hsa_circ_0074158 and QKI6 influence the proliferation, invasion, and migration of NSCLC cells. Circ_0074158 regulates PD-L1 ubiquitination by modulating QKI6 expression, thus affecting PD-L1 expression. Furthermore, circ_0074158 activates CD8⁺ T cells in PBMCs, inhibits immune escape, and promotes tumour cell apoptosis, suppressing tumour growth.

Conclusion

The circ_0074158/QKI6 axis regulates PD-L1 ubiquitination in NSCLC, limiting tumour cell proliferation and invasion. Our findings reveal a novel function for circ_0074158 in NSCLC, suggesting its potential as a therapeutic target.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00262-025-04152-7.

Introduction

Lung cancer, the leading cause of cancer-related mortality globally, is characterised by a dismal five-year survival rate of 15–20% [1]. Non-small cell lung cancer (NSCLC) accounts for approximately 85% of lung cancer cases [2, 3]. While radical resection can offer curative outcomes, the prognosis for NSCLC is frequently compromised due to the insidious progression of distant metastasis and recurrent local growth. This highlights the urgent need for the medical and scientific communities to identify novel molecular targets and biomarkers that can enhance both therapeutic and diagnostic strategies in the battle against lung cancer.

Circular RNAs (circRNAs), a distinct class of non-coding RNAs with a covalently closed-loop structure, have garnered significant attention in cancer research. This loop is formed through the ligation of the 3’ and 5’ ends by splicing factors, granting circRNAs remarkable stability against ribonuclease-mediated degradation [4]. In addition to their inherent stability, circRNAs are increasingly recognised for their potential as prognostic markers and therapeutic targets in oncology. They participate in a variety of cellular processes, interacting with RNA [5], DNA [6], and proteins [7]. Consequently, their dysregulation is linked to the pathogenesis of numerous cancers, including NSCLC, where they influence tumour growth and metastasis. For example, circSATB2 has been found to be upregulated in NSCLC, where it promotes the expression of fascin homologous protein 1 (FSCN1) via miR-326 [8]. Similarly, the overexpression of circ_0020123 in NSCLC is associated with enhanced cell proliferation and migration, along with reduced apoptosis, partly through the modulation of miR-590-5p [9]. Although current research has illuminated circRNAs’ role as miRNA sponges, the diverse mechanisms through which they influence NSCLC remain incompletely understood. A deeper exploration of circRNAs’ complex roles in NSCLC is essential to uncovering new insights into tumourigenesis and metastatic progression.

Quaking Homolog 6 (QKI6), a member of the Quaking RNA-binding protein (RBP) family, plays a pivotal role in RNA processing and regulation, influencing a wide array of biological processes [10]. Its overexpression has been shown to inhibit the proliferation and invasiveness of bladder cancer cells while simultaneously promoting apoptosis and cell cycle arrest [11]. In the context of NSCLC, elevated QKI6 expression is significantly linked to improved patient survival outcomes. Recent studies underscore the multifaceted role of QKI6 in tumourigenesis, demonstrating that its overexpression not only suppresses the proliferation and migration of NSCLC cells but also induces cell cycle arrest [12]. The protein homeostasis of PD-L1 is governed by multiple ubiquitination pathways. Recent studies have revealed that in colorectal cancer, ENO1 recruits the HECT/RBR-type E3 ligase STUB1 (CHIP) to catalyse K48-linked polyubiquitination of PD-L1, thereby promoting its degradation via both the 26S proteasome and the lysosomal pathway [13]. Glycosylation at ENO1-S249, however, disrupts the STUB1–PD-L1 interaction, suppresses ubiquitination, and stabilises PD-L1 [14]. Concurrently, FBW7 and FBXO38 enhance K48-linked polyubiquitination of PD-1 at Lys233 and its proteasomal degradation, further modulating the immune-checkpoint axis [15, 16]. Additionally, E3 ligases such as SPOP have been shown to directly ubiquitinate PD-L1, collectively forming a multilayered and reversible network that regulates tumour immune evasion [17].

In the present study, circ_0074158 can bind to QKI6, modulating the ubiquitination of PD-L1. Additionally, circ_0074158 inhibits tumour proliferation, invasion, and migration. It also activates CD8⁺ T cells, enhancing the expression of GzmB, perforin, IFN-γ, and TNF-α, thereby stimulating the tumour immune response in NSCLC.

Materials and methods

Cell culture and reagents

A549, 293 T, and Lewis lung carcinoma (LLC) cell lines were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Lentiviral vectors comprising four plasmids were used: vector plasmid PKG: PKGX, helper plasmid pMDLg/pRRE, helper plasmid pRSV-Rev, and helper plasmid pMD2. After plasmid transfection and collection of the viral supernatant, the lentivirus was concentrated, purified, and used to establish stable transient cell lines. Selection was performed using Puro (2 μg/mL).

For siRNA transfection, siRNA was first diluted in Opti-MEM (Thermo, 31985070). Then, 5 µL of Lipo3000 (Invitrogen, L3000015, USA) was diluted in 250 µL of Opti-MEM. The diluted siRNA and Lipo3000 were mixed and incubated, followed by addition to a cell plate containing cells in 500 μL of Opti-MEM medium. The mixture was incubated for 6 h. The sequences of the shRNA and siRNA are provided in Supplementary Table 1.

Real-time qPCR

RNA extraction was performed by adding 1 mL of pre-cooled TRIzol (Invitrogen, 15,596–026, USA) to the cells in a six-well plate, followed by the addition of 0.2 mL of chloroform. The samples were centrifuged at 12,000 × g at 4 °C for 15 min. After centrifugation, the supernatant was aspirated, and isopropanol and 1 mL of 70% ethanol were added. The supernatant was removed, and the RNA pellet was dissolved in 30 μL of RNase-free water. The RNA concentration was determined by measuring the absorbance at 260 and 280 nm using 5 μL of the RNA sample mixed with 495 μL of 1 × Tris–EDTA (TE) buffer. cDNA synthesis was performed using the cDNA First Strand Synthesis Kit (TaKaRa RR036B, Japan), with incubation at 37 °C for 15 min followed by 85 °C for 5 s. For amplification, TB Green® Premix Ex Taq™ II (Tli RNaseH Plus) (TaKaRa RR820A, Japan) was used, and GAPDH was employed as the housekeeping gene. Relative expression was calculated using the 2(ΔΔCt) method. All primer sequences are listed in Supplementary Table 2.

Western blot (WB) and immunoprecipitation (IP)

As previously described, cells were digested with an appropriate amount of trypsin at 37 °C and centrifuged [12]. To prepare the lysis buffer, 10 μL of phosphatase inhibitor, 1 μL of protease inhibitor, and 5 μL of 100 mM phenylmethylsulfonyl fluoride (PMSF) were added per millilitre of ice-cold buffer and mixed thoroughly. Protein quantification was performed using the bicinchoninic acid (BCA) assay. Protein samples were processed via SDS-PAGE and transferred to a polyvinylidene fluoride (PVDF) membrane. Blocking was done with milk for 2 h. Cells were incubated with the primary antibody at 4 °C overnight and with the secondary antibody at room temperature for 1 h, followed by visualisation using enhanced chemiluminescence (ECL). Imaging was carried out using the ChemiDoc™ Touch system (Bio-Rad). For IP, the following cell groups were used: A549 + OE-NC + HA-Ub + Flag-PD-L1 + MG132, A549 + OE-circRNA + HA-Ub + Flag-PD-L1 + MG132, A549 + OE-circRNA + si-NC + HA-Ub + Flag-PD-L1 + MG132, and A549 + OE-circRNA + si-QKI6 + HA-Ub + Flag-PD-L1 + MG132, where MG132 (MCE, HY-13259) acts as a proteasome inhibitor. The primary antibody (HA, PD-L1) or immunoprecipitated IgG was added and incubated overnight with gentle shaking. Next, 20 µL of fully resuspended Protein G Agarose (Biotin P2006) was added, and the mixture was gently shaken again. The supernatant was carefully aspirated. The following antibodies were used: QKI6 (Sigma, AB9906), STUB1 (Abcam, ab134064), Rabbit Anti-PD-L1 (Proteintech, 17952–1-AP), Mouse Anti-HA (Proteintech, 66006–2-Ig), Mouse Anti-PD-L1 (Proteintech, 66248–1-Ig), OTUB1 (Abcam, ab270959) and Rabbit Anti-GAPDH (KeyGEN BioTECH, KGC6102-1).

Flow cytometry

Cells in the logarithmic growth phase were digested and seeded into six-well plates. An appropriate culture medium was added based on group settings, and a negative control group was established. Following this, 1 × 10⁶ cells were collected, resuspended, and stained with PD-L1 (1:1000, ProteinTech, 17952–1-AP), CD3 (human)–fluorescein isothiocyanate (FITC) (BioLegend, 300406), and CD8 (human)–allophycocyanin (APC) (BioLegend, 344721).

Isoform control antibodies CD3 (human)–FITC (BioLegend, 400107) and CD8 (human)–APC (BioLegend, 400121) were incubated at 37 °C for 30 min in the dark. Tetramethylrhodamine (TRITC) fluorescent secondary antibody was added and incubated for an additional 30 min. Cell phenotypes were analysed using flow cytometry (BECKMAN COULTER CytoFLEX).

After 48 h of transfection treatments, cells entered the logarithmic growth phase. They were then digested, counted, and resuspended at a concentration of 1 × 10⁵ cells/mL for inoculation into six-well plates. The following day, once the cells adhered to the plate, 1 × 10⁶ cells were collected and suspended in 500 μL of binding buffer using the Annexin V-APC/7-AAD Apoptosis Detection Kit (China Jiangsu KGI Biotech Co., Ltd., KGA1106-50). Subsequently, 5 μL of Annexin V-APC was added to the cell suspension and mixed thoroughly, followed by the addition of 5 μL of 7-aminoactinomycin D (7-AAD). The mixture was incubated for 5–15 min, and apoptosis was analysed by flow cytometry.

RNA immunoprecipitation and high-throughput sequencing (RIP-seq) and rip-qPCR

A549 cells (5 × 10⁷) were centrifuged, and the cell lysate was incubated with the QKI6 antibody (Sigma, AB9906) to form a lysate-QKI6 antibody mixture. This mixture was then incubated with Protein G magnetic beads at 4 °C for 4 h. The final digested samples were mixed with 300 μL of triazole, followed by RNA purification using an RNA extraction kit. The purified RNA was utilised to construct a circRNA high-throughput sequencing library. The QKI6 protein was enriched with circRNA through RIP, and a library for high-throughput sequencing was created using the VAHTSTM Small RNA Library Prep Kit (Illumina). Both the 3’ and 5’ ends of the circRNA were ligated to a universal library at their respective ends, and connected to a universal connector. Subsequently, reverse transcription, PCR amplification, and magnetic bead purification were performed to obtain the sequencing library, which was sequenced on the Illumina® platform. To ensure high-quality RIP-seq data, RNA was extracted using TRIzol and treated with DNase I. Libraries were constructed using the NEBNext Ultra RNA Library Prep Kit, with size selection (150–200 bp) and quantification using Qubit and Agilent Bioanalyzer 2100. Sequencing was performed on an Illumina NovaSeq 6000 platform, generating 30–50 million reads per sample. Raw reads were filtered using FastQC and Trimmomatic to remove low-quality reads and contaminants, and only high-quality reads were used for analysis. Technical replicates were assessed for reproducibility (R² > 0.9).

Following the Magna RIP™ RNA-Binding Protein Immunoprecipitation Kit (cat. no. 17–700; Millipore) protocol, A549 cells (1 × 10⁷) were harvested and lysed in RIP lysis buffer containing proteinase K. Magnetic beads were prepared with 6 µL of QKI6 antibody (Sigma, AB9906) for immunoprecipitation. RNA–protein complexes were rotated at 4 °C overnight. RIP wash buffer containing 0.5 M EDTA and RNase was prepared; the complexes were captured with a magnetic stand and washed to remove unbound material. RNA was then extracted and analysed by qPCR. All primer sequences are listed in Supplementary Table 2.

Electrophoretic mobility shift assay (EMSA)

The cell samples were grouped as follows: 1) WT-labelled probe; 2) MUT-labelled probe; 3) A549 + WT-labelled probe + QKI antibody; 4) A549 + WT-labelled probe + 100 × WT unlabelled probe + QKI antibody (cold competition); 5) A549 + WT-labelled probe + 100 × MUT unlabelled probe + QKI antibody (cold competition); 6) A549 + MUT-labelled probe + QKI antibody; 7) A549 + MUT-labelled probe + 100 × WT unlabelled probe + QKI antibody (cold competition); 8) A549 + MUT-labelled probe + 100 × MUT unlabelled probe + QKI antibody (cold competition). Hsa_circ_0074158 was used as the probe.

The sequences for the WT and mutation probes are as follows:

WT probe: 5’-GAAGAUGUUCGAAAACAAGAAAUGACUGC-3’, 5’-GCAGUUCAUUUCUUUGUUUUUCGAACAUCUUC-3’.

Mutation probe: 5’-GAAGAUCAAAACAAGAAAUGACUGC-3’, 5’-GCAGUCAUUUCUUGUUUUGAUCUUC-3’.

For gel-shift analysis, the reagents were added in the order specified by the assay kit (China Jiangsu KGI Biotech Co., Ltd., KGS101). The mixture was incubated at room temperature for 10 min, followed by the addition of the labelled probes. To this mixture, 1 µL of EMSA/Gel-Shift Sampling Buffer (colourless, 10 ×) was added. The solution was mixed well, and a sample was immediately taken. The PVDF membrane was electrophoretically transferred and cross-linked for 45–60 s using a UV-light cross-linker (UV wavelength = 254 nm; energy = 120 mJ/cm²). After cross-linking, the biotin-labelled probe was detected via chemiluminescence.

CCK-8 assay

The samples were digested, counted, and prepared into a cell suspension with a concentration of 1 × 10⁵ cells/mL. 100 μL of the suspension was added to each well of a 96-well plate. After allowing the cells to adhere and grow for 24 h, CCK-8 staining was performed. The optical density (OD) was measured at 450 nm. The inhibition rate for each group was calculated using the following formula: Inhibition rate (%) = [(Negative control group—Experimental group) / Negative control group] × 100%.

Transwell migration assay

For migration assays, cells in the logarithmic growth phase were digested and seeded into chambers at a density of 5 × 10⁴ cells per 200 µL of serum-free medium per well. The lower chamber was filled with 600 µL of DMEM containing 10% foetal bovine serum. After 24 h of culture, the medium was removed, and the cells were washed twice with PBS, fixed with formaldehyde for 15 min, and stained with 0.1% crystal violet for 15 min. The cells were rinsed three times with distilled water, and images were taken under a microscope.

Transwell invasion assay

For invasion assays, Matrigel (BD Biosciences, USA) was thawed overnight at 4 °C. The next day, Matrigel was diluted 1:9 with serum-free DMEM, and 40 µL of the diluted Matrigel was added to each chamber. The chambers were incubated for 5 h. Cells were seeded as described in the migration assay. After 30 h, the medium was discarded, and the migrated cells were processed as described for the migration assay.

Wound healing assay

For wound healing assays, cells were seeded into six-well plates and scratched with a sterile tip. The plates were photographed under a microscope at 100 × magnification. The width of the scratch was measured at 0 h and 24 h.

Enzyme-linked immunosorbent assay (ELISA)

The original multiplex standards were diluted in EP tubes according to the instructions provided in the ELISA kits for TNF-α (mlbio, ml077385), IFN-γ (mlbio, ml077386), GzmB (mlbio, ml060098), and perforin (mlbio, ml063483). Blank, standard, and sample wells were prepared accordingly. Subsequently, 50 µL of the standard and 50 µL of the sample were added to each well of the ELISA plate, followed by 100 µL of the ELISA reagent. After shaking the plate to dry it, each well was filled with washing solution and left to stand for 30 s. The OD of each well was measured at 450 nm.

Co-culture of peripheral blood mononuclear cells (PBMCs) with tumour cells

Freshly collected leukocyte layers from human peripheral blood were transferred to a sterile 50-mL centrifuge tube. Pre-warmed PBS was added to the leukocyte layer at a 1:1 dilution. The cells were mixed thoroughly following the instructions from the Human Peripheral Blood Mononuclear Cell Isolation Solution Kit (China Solepol P8680). The resulting haematological cell suspension was centrifuged at room temperature, and the white mononuclear cell layer was gently aspirated and resuspended in PBS.

PBMCs were obtained as a cell suspension, and 200 μL of the suspension was added to each well in the upper chamber of a 24-well co-culture system. Depending on the experimental grouping, 800 μL of A549 cell suspension was added to the lower chamber of the co-culture system. The PBMCs were co-cultured with A549 cells for 24 h.

TUNEL assay

The co-cultured cells were fixed, and after permeabilisation, washing, and enzyme inactivation, a TUNEL assay kit (KGA1405-100, Jiangsu, China) and a DAPI staining kit (KGA1808-50, Jiangsu, China) were used. Then, 50 μL of the terminal deoxynucleotidyl transferase enzyme reaction solution was added and placed in a humidified chamber. Streptavidin-TRITC labelling solution (50 µL) was dropwise-added to each sample, which was then incubated in a humidified chamber in the dark. The samples were rinsed, sealed with a film, and colour development was performed for observation.

5-Ethynyl-2’-deoxyuridine (EdU) assay

A549 cells were retrieved from the lower chamber of the co-culture system and cultured using the key Fluor594 Click-iT EdU Imaging Kit (KGA333, Jiangsu, China). The cells were then fixed with 500 μL of a fixation solution and incubated for another 30 min. Following this, 200 μL of 2 mg/mL glycine was added, and the samples were shaken for 5 min. Subsequently, 500 μL of a staining reaction solution was added to each well, and the cells were incubated for 30 min on a shaker. Afterwards, 500 μL of an osmotic agent (0.5% TritonX-100 in PBS) was added, followed by the addition of 500 μL of the reaction solution. The cells were incubated for 15 min, and fluorescence microscope imaging was performed at a magnification of 200 × .

Animal and model

Cultured mouse lung cancer LLC cells were collected at a concentration of 1 × 10⁸ cells/mL. 0.1 mL of this suspension was inoculated subcutaneously into the dorsal shoulder of the right forelimb of each 4-week-old C57 mouse. The diameters of the transplanted tumours were measured using Vernier callipers. Once the tumours reached 100 mm³, both animal weight and tumour volume (TV) were recorded every 3 days. At the end of the experiment, the tumours were surgically removed and weighed. TV was calculated using the formula: TV = 1/2 × a × b², where a represents the length and b represents the width.

Haematoxylin & eosin staining (H&E) and Immunohistochemistry (IHC)

The samples were treated following the protocol described in a previous study [18]. After processing, the samples were stained with an H&E staining solution kit (China Ltd. KGA224) and examined microscopically.

Paraffin sections were processed as described previously [18]. Primary antibody CD8 (1:500, Proteintech 29896–1-AP) was applied overnight at 4 °C. The biotin-labelled secondary antibody was added, and the sections were incubated for 20 min at room temperature. DAB solution was applied to each section, followed by the termination of the chromogenic reaction. The sections were then immersed in a haematoxylin staining solution. After dehydration and sealing, the slices were scanned using a digital pathology slice scanner.

Statistical analysis

Statistical analysis was performed using GraphPad Prism software (version 8.0). A two-tailed Student’s t test was used to assess statistical significance between two groups. Each experiment was repeated three times to ensure accuracy. Statistical significance was set at P < 0.05.

Results

QKI6 increased PD-L1 ubiquitination levels in NSCLC

A QKI6-overexpression A549 cell line was established, and siRNA-mediated silencing of QKI6 led to a significant reduction in QKI6 expression in all three siRNA knockdowns (P < 0.001). The most effective knockdown was achieved with siRNA-1 (Fig. 1A). Therefore, siRNA-1 was selected for use in subsequent experiments. qPCR analysis revealed no significant change in PD-L1 RNA levels in the A549 cell line, regardless of QKI6 overexpression or knockdown (Fig. 1B). However, flow cytometry showed a significant reduction in PD-L1 expression in QKI6-overexpressing A549 cells, while PD-L1 expression increased following QKI6 knockdown (Fig. 1C and D, P < 0.01). Protein analysis revealed a significant decrease in PD-L1 protein levels in QKI6-overexpressing cells, while PD-L1 expression was significantly elevated in QKI6-knockdown cells (Fig. 1E and F, P < 0.05). Despite QKI6 overexpression not affecting PD-L1 RNA levels, QKI6 overexpression enhanced the ubiquitination of PD-L1, which, in turn, reduced its protein levels, as demonstrated by Co-IP experiments (Fig. 1G). Western blot analysis revealed that STUB1 protein levels were markedly elevated upon QKI6 overexpression, whereas STUB1 expression was significantly reduced when QKI6 was silenced (Fig. 1H).

Fig. 1.

QKI6 modulates the ubiquitination levels of PD-L1 in NSCLC. A–B: qPCR analysis of A QKI6 expression and B PD-L1 expression in A549 cells, QKI6-overexpressing A549 cells, and QKI6-knockdown A549 cells. C and D Flow cytometry analysis for detecting the phenotypic expression of PD-L1 in A549 QKI6-overexpressing and QKI6-knockdown cell lines. E and F WB assay for PD-L1 expression in A549 cells, QKI6-overexpressing A549 cells, and QKI6-knockdown A549 cells. G PD-L1 ubiquitination analysis was performed on different cell fractions using Co-IP. The following cell groupings were used: (1) A549 + overexpressing NC + HA-Ub + Flag-PD-L1; (2) A549 + overexpressing NC + HA-Ub + Flag-PD-L1 + MG132; (3) A549 + overexpressing QKI6 + HA-Ub + Flag-PD-L1 + MG132. H WB assay for STUB1 expression in A549 cells, QKI6-overexpressing A549 cells, and QKI6-knockdown A549 cells

Combination of hsa_circ_0074158 and QKI6

To identify circRNAs binding to QKI6, a QKI6 antibody was used to enrich QKI6-bound circRNAs in A549 cells. These circRNAs were further analysed through RIP-seq. RIP-circRNA sequencing identified 2,389 circRNAs (Supplementary Table 3). Compared to the input IP, 285 circRNAs were significantly enriched by IP (Fold Change > 2) (Fig. 2A). These circRNAs were annotated in the circBase database (http://www.circbase.org/), and their homology and similarity to mouse sources were also compared. Among the 285 enriched circRNAs, three circRNAs (circ_0074158, circ_0001470, and circ_0099287) were selected based on their higher homology to mouse circRNAs and their higher predicted binding scores in the RBPsuite database (Supplementary Table 4). qPCR analysis showed that circ_0074158 was highly expressed in QKI6-overexpressing A549 cells (P < 0.01) and expressed at lower levels in QKI6-knockdown cells (P < 0.05) (Fig. 2B). Circ_0099287 expression decreased in the QKI6-overexpressing lines (Fig. 2C), while circ_0001470 did not show differential expression in the QKI6-knockdown cells (Fig. 2D). Thus, hsa_circ_0074158 was selected for further study. Hsa_circ_0074158 is formed via reverse splicing from CTNNA1 exons 2 and 3 (Fig. 2E). Gene ontology (GO) enrichment analysis revealed that circ_0074158 was primarily involved in DNA damage regulation, histone modification, chromosomal regions, and Ras guanylate exchange factor activity (Fig. 2F). KEGG pathway analysis highlighted its involvement in the cell cycle, ubiquitin-mediated protein degradation, and cellular senescence, as well as its roles in metabolism and cancer (Fig. 2G and H). RIP-qPCR assays confirmed the direct interaction between hsa-circ_0074158 and QKI6 (Fig. 2I). Further evidence from electrophoretic mobility shift assays (EMSA) confirmed that QKI6 binds to hsa_circ_0074158 (Fig. 2J).

Fig. 2.

Interaction between hsa_circ_0074158 and QKI6. A CircRNAs binding to QKI6 in A549 cells were detected by RIP-seq. A total of 284 circRNAs showed increased expression, while 1951 circRNAs exhibited decreased expression. B–D Expression levels of hsa_circ_0074158, hsa_circ_0001470, and hsa_circ_0099287 were measured by qPCR. Hsa_circ_0074158 was significantly more expressed in QKI6 high-expression cell lines (P < 0.001), and its levels were reduced in QKI6-knockdown cell lines (P < 0.01). E Schematic illustrating the mechanism of hsa_circ_0074158. F GO enrichment analysis of hsa_circ_0074158. G and H KEGG enrichment analysis of hsa_circ_0074158. I rip-qPCR results of circ_0074158. J EMSA confirming that QKI6 binds to hsa_circ_0074158

Circ_0074158 inhibited proliferation, invasion, and migration in NSCLC

Circ_0074158-overexpressing and knockdown A549 cell lines were established, with qPCR confirming the presence of circ_0074158 and validating the formation of circRNAs (Fig. 3A and B). Overexpression of circ_0074158 significantly inhibited cell proliferation, whereas its knockdown promoted cell growth. Additionally, QKI6 knockdown in circ_0074158-overexpressing cells led to a reduced cell growth rate (Fig. 3C). Flow cytometry analysis revealed a significant increase in apoptosis rates in circ_0074158-overexpressing A549 cells (P < 0.001). In contrast, the knockdown of QKI6 in circ_0074158-overexpressing A549 cells resulted in decreased apoptosis levels (Fig. 3D and E). Overexpression of circ_0074158 decreased cell invasion (P < 0.01), while its knockdown enhanced cell invasion (P < 0.01). Furthermore, the knockdown of QKI6 in circ_0074158-overexpressing cells significantly increased their invasive and migratory capacities (Fig. 3F and G, P < 0.0001). Similar results were observed in the wound healing assay, where circ_0074158 overexpression reduced cell migration, while its knockdown enhanced it (Fig. 3H and I, P < 0.01). WB revealed that E-cadherin expression was upregulated in circ_0074158-overexpressing cells, while N-cadherin expression was downregulated. In contrast, the shcirc_0074158 group exhibited reversed expression patterns for both proteins (Fig. 3J).

Fig. 3.

Circ_0074158 overexpression reduces proliferation, invasion, and migration in NSCLC. A and B qPCR assay for hsa_circ_0074158 expression in A549 cell lines with knockdown and overexpression. A549-shRNA-1 showed the most effective knockdown efficiency (P < 0.0001). C Cell proliferation inhibition of A549 + circ_0074158 was significantly greater than the control (P < 0.0001). The proliferation inhibition rate of A549 + shcirc_0074158 was significantly lower than the control (P < 0.0001). Low expression of QKI6 significantly enhances cell proliferation. High expression of QKI6 increased the inhibition rate of cell proliferation (P < 0.001). D and E Apoptosis level of A549 + circ_0074158 was significantly higher than the control in flow cytometry analysis (P < 0.0001). High expression of QKI6 increased apoptosis levels (P < 0.0001). F and G Cell invasion and migration ability of A549 + circ_0074158 was significantly lower than the control (P < 0.01), whereas A549 + shcirc_0074158 had significantly higher invasion and migration ability (P < 0.01). High expression of QKI6 reduced cell invasion and migration (P < 0.0001). H and I Cell migration ability assessed by wound healing assay in A549 + circ_0074158 was significantly lower than the control (P < 0.0001), whereas A549 + shcirc_0074158 exhibited significantly higher migration (P < 0.01). High expression of QKI6 reduced cell migration (P < 0.01). J WB detection of E-cadherin and N-cadherin expression in A549, A549 + circ_0074158, and A549 + shcirc_0074158 cells

Circ_0074158 increased the ubiquitination level of PD-L1

The RNA expression levels of PD-L1 and QKI6 were assessed across different cell lines using qPCR. No significant changes in PD-L1 RNA levels were observed, regardless of whether circ_0074158 was overexpressed or QKI6 was knocked down (Fig. 4A). However, protein analysis revealed that QKI6 levels significantly increased following circ_0074158 overexpression, while QKI6 was significantly downregulated after circ_0074158 knockdown. In contrast, the protein levels of PD-L1 displayed an opposing trend. Specifically, the knockdown of QKI6 in A549 cells, in conjunction with circ_0074158 overexpression, led to an increase in PD-L1 protein levels (Fig. 4B and C). Flow cytometry confirmed a significant reduction in PD-L1 expression after circ_0074158 overexpression, whereas circ_0074158 knockdown resulted in the opposite effect. Additionally, QKI6 knockdown in circ_0074158-overexpressing cells significantly increased PD-L1 levels (Fig. 4D and E). Co-IP experiments further demonstrated that circ_0074158, in association with QKI6, enhanced PD-L1 ubiquitination levels (Fig. 4F).

Fig. 4.

circ_0074158 enhances the ubiquitination level of PD-L1 in association with QKI6. A qPCR analysis revealed no significant change in the RNA expression levels of PD-L1 across different cell lines. The RNA level of QKI6 was overexpressed in circ_0074158-high-expressing cell lines (P < 0.001) and reduced in circ_0074158-knockdown cell lines (P < 0.001). B and C The protein level of PD-L1 was significantly lower in circ_0074158-high-expressing cell lines (P < 0.001) and higher in circ_0074158-knockdown cell lines. PD-L1 expression was also decreased in QKI6 high-expressing cell lines (P < 0.001). The protein level of QKI6 was overexpressed in circ_0074158-high-expressing cell lines (P < 0.001) and reduced in circ_0074158-knockdown cell lines (P < 0.01). D and E Flow cytometry analysis revealed a decrease in PD-L1 expression in circ_0074158-high-expressing cell lines (P < 0.01), while expression increased in the circ_0074158-knockdown cell lines (P < 0.001). Knockdown of QKI6 increased PD-L1 expression (P < 0.001). F Co-IP assay for PD-L1 ubiquitination levels. High circ_0074158 expression significantly increased PD-L1 ubiquitination. Knockdown of QKI6 also enhanced PD-L1 ubiquitination

CircRNA activated CD8⁺T cells in PBMCs and inhibited immune escape

To explore the role of circ_0074158 in the tumour immune microenvironment (TIME), PBMCs were co-cultured with various cell lines, which were classified into four groups: (a) A549 + overexpressing NC/PBMC, (b) A549 + overexpressing circRNA/PBMC, (c) A549 + sh-NC/PBMC, and (d) A549 + sh-circRNA/PBMC. The expression of GzmB, perforin, IFN-γ, and TNF-α in these groups, influenced by circ_0074158 overexpression, was analysed via ELISA, showing elevated levels in the overexpressing group (Fig. 5A, B, C and D), with a corresponding reduction in these markers observed in the circ_0074158-knockdown cells. Cell proliferation was assessed using the EdU assay, revealing decreased proliferation in circ_0074158-overexpressing cells and increased proliferation in the circ_0074158-knockdown group (Fig. 5E and G). TUNEL staining was performed to detect apoptosis, showing a significant increase in apoptotic cells in the circ_0074158-overexpression group, while the circ_0074158-knockdown cells exhibited significantly lower apoptosis levels (Fig. 5F and H). Additionally, a significant increase in the number of CD8⁺ T lymphocytes was observed in the circ_0074158-overexpressing group, whereas the circ_0074158-knockdown group showed a marked decrease in CD8⁺ T cell numbers (Fig. 5I and J).

Fig. 5.

CircRNA activates CD8⁺ T cells in PBMCs and inhibits immune escape. A–D PBMCs were co-cultured with A549 + NC cells, A549 + circ_0074158-overexpressing cells, A549 + sh-NC cells, and A549 + circ_0074158-knockdown cells. ELISA assays for granzyme B (Gra B) (A), Perforin (B), IFN-γ (C), and TNF-α (D) secretion were significantly increased in the circ_0074158-overexpressing group (P < 0.001) and significantly inhibited in the circ_0074158-knockdown group (P < 0.001). E, G: Circ_0074158 overexpression inhibited cell proliferation as detected by EdU assays (E), while its knockdown restored cell proliferation ability (G) (P < 0.0001). F, H TUNEL assays showed that apoptosis levels in the circ_0074158-overexpression group were significantly higher than in the control group (P < 0.0001), while apoptosis levels were significantly lower in the circ_0074158-knockdown group (P < 0.0001). I and J Flow cytometry analysis of the co-culture system revealed that the ratio of CD3⁺CD8⁺ T cells was significantly higher in the circ_0074158-overexpression group compared to the control (P < 0.001), whereas the opposite was observed for the circ_0074158-knockdown group (P < 0.01)

In vivo experiments demonstrate that circRNA inhibits tumours

Mouse-derived mmu_circ_0000860 shares approximately 95% homology with hsa_circ_0074158 (Supplementary Fig. 1 and Supplementary Table 5). A schematic of mmu_circ_0000860 formation is shown in Fig. 6A. Mouse-derived mmu_circ_0000860-overexpressing LLC cell lines were constructed, and expression levels of mmu_circ_0000860 were measured via qPCR (Fig. 6B). Overexpression of mmu_circ_0000860 significantly inhibited subcutaneous tumour growth, with TV in C57 mice inoculated with mmu_circ_0000860-overexpressing cells markedly reduced (Fig. 6C, D, E, and F). Cells from the control group were disorganised, irregularly arranged, densely packed, and exhibited varying cell sizes and enlarged nuclei. In contrast, cells from the mmu_circ_0000860-overexpression group displayed varying degrees of crumpling, dense cytoplasm, nuclear chromatin condensation, nuclear consolidation (indicated by red arrows), as well as signs of denaturation and lysis (indicated by red circles). Necrotic areas are highlighted in pink (Fig. 6G). IHC analysis of mouse subcutaneous tumour tissues revealed significantly higher CD8 expression in tumours from mmu_circ_0000860-overexpressing mice compared to the control group (Fig. 6H and I). qPCR analysis of tumour tissues showed significant increases in the expression of mmu_circ_0000860 (Fig. 6J), PD-1 (Fig. 6K), and QKI6 (Fig. 6M), while PD-L1 expression (Fig. 6L) remained unchanged. Consistent results were obtained in cellular experiments. Furthermore, protein analysis of QKI6, PD-L1, STUB1, and OTUB1 in three pairs of tumour tissues revealed significant upregulation of QKI6 and STUB1 in mmu_circ_0000860-overexpressing tissues, while PD-L1 and OTUB1 levels were notably reduced (Fig. 6N).

Fig. 6.

Murine-derived mmu_circ_0000860, a homologue of circ_0074158, inhibits tumours in vivo. A Schematic representation of the constructed murine-derived mmu_circ_0000860. B qPCR verification showing that the expression of mmu_circ_0000860 was significantly higher than that in the control group (P < 0.0001). C Subcutaneous tumours in C57 mice were injected with two different cell lines, LLC + NC and LLC + mmu_circ_0000860. D Dissection of the subcutaneous tumours shown in (C). E Subcutaneous tumour volume in C57 mice with mmu_circ_0000860 overexpression was significantly smaller than that in the control group (P < 0.0001). F Subcutaneous tumour weight in C57 mice with mmu_circ_0000860 overexpression was significantly lower than that of the control group (P < 0.0001). G Haematoxylin and eosin (H&E) staining of LLC + NC and LLC + mmu_circ_0000860 subcutaneous tumours (100 × and 400 × magnification, respectively). H IHC staining to detect CD8 expression in subcutaneous tumour tissues of LLC + NC and LLC + mmu_circ_0000860 (200 μm and 50 μm, respectively). I CD8 IHC score analysis showing that CD8 expression in tissues of LLC + NC was lower than in the tissue of LLC + mmu_circ_0000860. J–M: qPCR analysis showing that the expression levels of mmu_circ_0000860, PD-1, and QKI6 RNA were significantly higher in the subcutaneous tumours of LLC + mmu_circ_0000860 compared to the control group (P < 0.0001). However, PD-L1 expression did not change significantly. N Protein expression levels of PD-1 and OTUB1 were significantly lower than in the control group in three pairs of subcutaneous tumours of LLC + mmu_circ_0000860, while QKI6 and STUB1 levels were significantly higher. O Mechanism diagram illustrating how circ_0074158/QKI6 promotes PD-L1 ubiquitination, activates CD8⁺ T cells, and suppresses tumour immune evasion

Discussion

Despite significant advancements in treatment options for NSCLC, the prognosis for patients remains poor. Therefore, extensive research is essential to identify novel diagnostic and therapeutic molecular markers. CircRNAs, due to their high stability, are emerging as ideal biomarkers for various diseases, including cancer [19, 20]. With ongoing research into cancer-associated circRNAs, numerous circRNAs have been identified in relation to tumourigenesis in NSCLC [21]. The present study identified circ_0074158, which targets and regulates QKI6, influencing the ubiquitination of PD-L1 in NSCLC cells and inhibiting tumour cell proliferation and invasion. Additionally, circ_0074158 activated CD8⁺ T cells in PBMCs, helping prevent tumour immune escape (Fig. 6O). These findings were further validated through a series of in vivo experiments.

While many studies have explored the role of circRNAs as miRNA sponges, where they bind to specific miRNAs to release RBPs and regulate downstream gene expression, few studies have directly investigated how circRNAs impact NSCLC through binding with RBPs. High-throughput sequencing and qPCR validation revealed a positive correlation between hsa_circ_0074158 and QKI6 expression levels, with QKI6 directly binding to hsa_circ_0074158. This interaction may regulate gene expression at the post-transcriptional level, thereby affecting the biological behaviour of NSCLC cells. KEGG and GO enrichment analyses indicated that the parental gene of circ_0074158, CTNNA1, is primarily involved in ubiquitin-mediated protein degradation and cellular senescence, processes essential to metabolism and cancer, aligning with key mechanisms of tumour development.circ_0074158/QKI6-mediated ubiquitination not only directly promotes proteasomal degradation but may indirectly influence lysosomal flux of PD-L1; this can be further examined with lysosomal inhibitors in future work.

CTNNA1, also known as α-E-catenin, plays a critical role in cell adhesion junctions and signal transduction, suppressing proliferation and epithelial–mesenchymal transition (EMT) in various tumour cells [22–24]. However, circCTNNA1 (hsa_circ_0074169) has been shown to promote the proliferation and invasion of colon cancer cells via the circCTNNA1/miR-149-5p/FOXM1 axis [25]. Furthermore, as a competitive endogenous RNA (ceRNA) for miR-363-3p, hsa_circ_0074169 enhances CXCL5 expression, promoting the proliferation and metastasis of colorectal cancer cells [26]. Thus, the role of circCTNNA1 in tumours warrants further investigation. In the present study, circ_0074158 regulated tumour cell proliferation and invasion by influencing QKI6 and PD-L1 in NSCLC. Additionally, the overexpression or knockdown of circ_0074158/QKI6 significantly altered PD-L1 protein levels without affecting PD-L1 RNA levels. This suggests that circ_0074158/QKI6 may regulate PD-L1 expression through a protein degradation mechanism.

As an immune-checkpoint molecule, the expression level of PD-L1 on tumour cell surfaces is closely linked to tumour immune evasion [27]. The circRNA hsa_circ_0136666, which is highly expressed in gastric cancer, plays a pivotal role in promoting metastasis and facilitating immune escape. It regulates the phosphorylation of PD-L1 via the protein kinase DNAPK through the hsa_circ_0136666/miR-375/PRKDC signalling pathway, inhibiting PD-L1 degradation. This leads to abnormal PD-L1 accumulation, thereby triggering immune escape [28]. OTUB1, a deubiquitinating enzyme, specifically binds to the intracellular domain (ICD) of PD-L1 and removes Lys-48-linked polyubiquitin chains through its deubiquitinating activity, preventing PD-L1 degradation in the endoplasmic reticulum. Depletion of OTUB1 results in reduced PD-L1 protein levels and diminished PD-L1 binding to PD-1 on tumour cells, while enhancing tumour cell sensitivity to immune cell-mediated cytotoxicity [29]. STUB1, an E3 ubiquitin ligase, may regulate PD-L1 expression by modulating the response of tumour cells to IFN-γ [30]. Our findings suggest that circ_0074158/QKI6 promotes PD-L1 degradation by increasing its ubiquitination, a mechanism that may contribute to the circ_0074158/QKI6-induced inhibition of immune escape in tumour cells. Future Co-IP/MS experiments will be conducted to systematically screen for E3 ubiquitin ligases (e.g. STUB1) that associate with QKI6, thereby clarifying the precise molecular mechanism by which QKI6 promotes PD-L1 ubiquitination.

CD8⁺ T cells are cytotoxic T lymphocytes critical for eliminating tumour cells [31]. The overexpression of circCCAR1 promotes PD-L1 transcription by binding to β-catenin, which inhibits CD8⁺ T cell functions and supports tumour growth [32]. In lung adenocarcinoma, circRNA-002178 has been shown to promote PD-L1 expression. CircRNA-002178 is delivered to CD8⁺ T cells via exosomes, enhancing PD-1 expression and facilitating immune evasion by tumour cells [33]. Our findings suggest that circ_0074158 promotes the activation and proliferation of CD8⁺ T cells in PBMCs, enhancing their ability to kill tumour cells, thereby inhibiting tumour progression and strengthening immune surveillance. Through this mechanism, circ_0074158 may facilitate immune system recognition and clearance of tumours, reducing the likelihood of immune escape. In vivo experiments further validate the function of circ_0074158. Observing tumour growth rate, immune cell infiltration, and activity in an animal model will help confirm its role in tumour development and immune response. Circ_0074158/QKI6 lowers PD-L1 and boosts CD8⁺ T cell infiltration; combining this axis with PD-1/PD-L1 blockade may therefore improve response rates and delay resistance. Due to experimental constraints and sample size limitations, this study failed to isolate CD8⁺ T cells for specific co-culture or perform flow cytometry with cell fixation and cytokine staining. To address these limitations, future experiments will involve isolating CD8⁺ T cells from PBMCs using magnetic bead sorting and co-culturing them with cell lines. Additionally, the role of the circ_0074158/QKI6 axis in tumour cell-CD8⁺ T cell interactions will be explored using flow cytometry, multi-marker immunofluorescence, and ELISA.

This study highlights the critical roles of QKI6 and hsa_circ_0074158 in NSCLC, particularly in the regulation of PD-L1 expression and tumour immune escape. These findings may offer valuable insights for developing new therapeutic strategies for NSCLC and aid in designing personalised therapies targeting specific molecular pathways.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization:Y.N., X.Y., J.W., and H.Z. Datacuration:H.Z., X.L., J.L., and F.T. Formal analysis:H.Z., X.L., and F.L. Methodology:H.Z., K.L., and S.K. Validation: J.W., X.Y., and Y.N. Visualization:X.Y., and Y.N. Writing- original draft: H.Z., X.L., and X.L. Writing- review and editing: J.W., X.Y., and Y.N. Funding acquisition: Y.N. Project administration: Y.N. Supervision: Y.N., and J.W. All authors have read and agreed to the published version of the manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (No. 81301989) to Yunfeng Ni and the Key Research and Development Programme of Shaanxi (No. 2021SF-102).

Data availability

The data supporting the findings of this study are included within the article.

Declarations

Conflict of interest

The authors declare no competing interests.

Ethics approval

The nude mouse experiments were conducted in accordance with the necessary ethical standards and received approval from the Animal Ethics Committee of Tangdu Hospital, Fourth Military Medical University.