Background
Metastasis is the leading cause of cancer-related mortality and consists of multistep processes driven by the cooperation of cancer cells with the tumor microenvironment (TME) [
1]. As predominant stromal components in the TME, cancer-associated fibroblasts (CAFs), also known as activated fibroblasts, actively orchestrate a supportive niche as “fertilized soil” and endow incipient cancer cells with traits needed to develop metastases [
2]. CAFs are also identified as a generalized biomarker of the fibrotic TME subtype that correlates with inferior therapeutic outcomes and prognosis across various cancers [
3]. With the development of single-cell sequencing approaches, tremendous insight has been yielded into the heterogeneity and plasticity of CAFs. The composition and functional states of fibroblasts differ extensively in the evolving TME, which makes targeting CAFs challenging in clinical settings [
4‐
6]. Therefore, reprogramming CAFs into a “normal” or quiescent state might be a promising approach that benefits early cancer treatment and inhibits metastasis.
Accumulating evidence has implied that aberrant fibroblast activation is an early event before cancer cell dissemination [
7]. When the interaction with cancer cells begins, local normal fibroblasts (NFs) are usually among the first cell types to be recruited and activated into CAFs, and ultimately establish a niche to facilitate metastatic cascades [
8,
9]. Previous studies have shown that paracrine signaling molecules drive fibroblasts from homeostasis to activated states, including transforming growth factor β (TGF-β), platelet-derived growth factor (PDGF), fibroblast growth factor (FGF), and interleukin-1 [
7]. Moreover, direct physical contact with cancer cells via ligand receptor binding also leads to the transition of NFs into CAFs [
10]. More interestingly, CAFs maintain activated phenotypes even in the absence of cancer cells [
11]. However, the underlying mechanisms of fibroblast activation and phenotype switching that drive metastasis have not been extensively elucidated.
Fibroblast activation protein (FAP), a type II membrane serine protease, is widely acknowledged as one of the canonical CAF markers [
12]. FAP is overexpressed on activated fibroblasts and coincides with poor prognosis in cancers, which has made FAP-targeted imaging and therapy appealing [
13]. Indeed, over 28 different cancer entities have been accurately diagnosed in patients utilizing FAP ligands, including metastatic lesions [
14]. Cumulative studies have revealed that FAP-positive CAFs (FAP
+CAFs) account for the main phenotypes of activated fibroblasts in the TME, and they execute crucial functions in promoting tumor growth, angiogenesis and metastasis, as well as the formation and maintenance of an immunosuppressive microenvironment [
15‐
17]. In addition, our previous work demonstrated that FAP orchestrated the immune microenvironment and served as a biomarker of immunotherapy resistance. Patients with high FAP expression had a significantly shorter progression-free survival in immunotherapy of non-small cell lung cancer (NSCLC) [
18]. However, the upstream epigenetic mechanisms by which FAP drives tumor progression are unclear and warrant further investigation.
In this study, we determined that FAP+CAFs are correlated with metastasis and poor prognosis in NSCLC patients. By investigating the critical mediator in fibroblast activation, we identified a novel circular RNA (circRNA), circNOX4, that is significantly upregulated in CAFs and correlates with the worse overall survival of NSCLC patients. Importantly, circNOX4 fuels tumor growth and metastasis by activating the fibroblast niche via the miR-329-5p/FAP/IL-6 axis. Our results elucidate a novel mechanism of fibroblast activation and IL-6 signaling by CAFs to establish an inflammatory niche and highlight circNOX4 as a promising candidate for CAF-targeted therapy in NSCLC.
Materials and methods
Patients and clinical samples
Tumor samples were collected from a cohort of 84 NSCLC patients at the Fourth Hospital of Hebei Medical University (Shijiazhuang, China) between August 2018 and March 2019 with the approval of the Medical Ethics Committee and informed consent from the corresponding participants (2018MEC005). All samples were histologically confirmed to have NSCLC. Patients’ clinical information was collected and stored in a database, which was updated every 3 months by telephone follow-up. Overall survival (OS) was defined as the time interval from diagnosis to the date of cancer-related death or the last follow-up (January 2024).
Immunohistochemistry (IHC)
IHC staining was performed according to standard protocols as previously described [
18]. The staining was evaluated based on the intensity of FAP (ab207178, 1:250, Abcam, CA, USA) by two independent pathologists, and all scoring was completed blinded to the clinical information. The intensity of FAP was classified as 0, no staining; 1, weak intensity; 2, moderate intensity; and 3, high intensity. Scores of 0 and 1 were considered low expression, while scores of 2 and 3 were considered high expression.
Isolation, identification, and culture of fibroblasts from human NSCLC tissues
Five paired CAFs and NFs were isolated from human NSCLC tissues and normal tissues (more than 5 cm away from the tumor margin). Freshly isolated surgical specimens were acquired from patients with the approval of The Fourth Hospital of Hebei Medical University (2018MEC005). The patient’s basic information is listed in Additional file: Table
S1. For fibroblast isolation, tissue was initially stored and washed in PBS with penicillin/streptomycin (100 U/100 μg, BI, Israel). The tissue was then physically minced and digested enzymatically in a solution containing 2 mg/mL DNase (Gibco, Waltham, MA, USA), 5 mg/mL collagenase type I (Gibco, Waltham, MA, USA), 100 U/mL penicillin, and 100 μg/mL streptomycin at 37 °C for 2 h. After a series of filtration and centrifugation steps, cells were subsequently cultured in 15% fetal bovine serum (FBS, Gibco, Waltham, MA, USA), DMEM (Gibco, Waltham, MA, USA), 20 ng/mL EGF (PeproTech, Suzhou, China) and penicillin/streptomycin (100 U/100 μg) at 37 °C in humidified air with 5% CO
2 for subsequent expansion. Fibroblast populations were identified by morphology, western blotting and immunofluorescence staining (fibroblast markers α-SMA and FAP; epithelial cell marker EpCAM). Fibroblasts were subcultured at 80% confluence and used in experiments at passages 4–10. Antibody information is provided in Additional file: Table
S2.
Cell lines
Human lung cancer cell lines (A549, PC9, H226, and H1581) and human bronchial epithelial cells (BEAS-2B) were purchased from Procell Life Science & Technology Co., Ltd. (Wuhan, China) and maintained in RPMI-1640 medium (Gibco, Waltham, MA, USA) containing 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin at 37 °C and 5% CO2 under saturated humidity.
circRNA microarray
The circRNA microarray analysis (GEO accession number: GSE244065) was performed by the Sinotech Genomics Corporation, Shanghai, China. In brief, total RNA from three paired sets of CAFs and NFs was extracted by TRIzol reagent (Life Technologies, Carlsbad, CA, US) and purified using the RNeasy Mini Kit (Qiagen, GmBH, Germany). Then, RNA samples were utilized to generate biotinylated cRNA targets for the Sino Human ceRNA array V3.0. After hybridization, slides were screened by the Agilent Microarray Scanner (Agilent Technologies, Santa Clara, CA, US). Raw data were extracted and normalized, and circRNAs with a fold change of at least 1.5 and P values < 0.05 were selected for further analysis by R software.
Confirmation of the circular structure
Both complementary DNA (cDNA) and genomic DNA (gDNA) extracted from CAFs and NFs were amplified with circNOX4 primers (both convergent and divergent primers). Agarose gel electrophoresis was applied to analyze the RT–PCR products. The back-splice junction site of circNOX4 was verified by Sanger sequencing. Extracted RNAs were treated with RNase R (Geneseed, 3 U/μg, 37 °C, 20 min) to detect the stability of circRNA.
RNA isolation, cDNA synthesis, and quantitative RT-PCR (qRT-PCR) assay
In brief, total RNA was isolated using the E.Z.N.A. Total RNA Kit (OMEGA, Japan). cDNA was synthesized using HiScript III RT SuperMix (Vazyme, Nanjing, China) and amplified using a Light Cycler 480 II Real-Time PCR System (Roche, Basel, Switzerland) with ChamQ Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China). GAPDH and U6 were used as internal controls via the classical ΔΔCt method. Primers were designed and synthesized by RiboBio (Guangzhou, China) and Tsingke (Wuhan, China). The primer sequences are listed in Additional file: Table
S3.
RNA-fluorescence in situ hybridization (FISH)
FISH assays were performed on CAFs/NFs and NSCLC tissues as previously described [
19]. FISH probes were designed and synthesized by Servicebio (Wuhan, China). Dig-labeled probes specific for circNOX4 and biotinylated locked nucleic acid miR-329-5p probes were used during hybridization. The targeted sequences of the probes are provided in Additional file: Table
S4. The FISH score for circNOX4 in NSCLC tissues was determined based on the integrated optical density values scanned by AIPATHWELL software (Servicebio Technology Co., Wuhan, China). The optimal cut-off value of 0.1 was determined by X-tile software. Based on this threshold value, patients were categorized into circNOX4 high-expression or low-expression subgroups for further analysis.
Collagen contraction assay
The fibroblasts were embedded in three-dimensional collagen matrices in 24-well plates with Rat Tail Collagen I (Corning, NY, USA) at a final concentration of 2 mg/mL collagen and 5 × 105 cells per mL. A 500 μL aliquot of this mixture was dispensed into each well and allowed to solidify for 30 min at 37 °C. Gels were freed from the wells using a pipette tip after polymerization, and complete culture medium was added to the wells, followed by incubation for 12 h. ImageJ software was employed to measure the contractile capacity, which is defined by the following equation: [Area (well)-Area (gel)/Area (well)] × 100%.
Western blotting analysis
Standard western blotting was carried out by a wet transfer system (Bio-Rad Laboratories). Briefly, cells were lysed in lysis buffer (Solarbio, Beijing, China) supplemented with protease inhibitors. Total proteins were separated on SDS–PAGE gels and then transferred onto PVDF membranes (Millipore, Billerica, MA, USA). After blocking the membranes with 5% milk-TBST, they were incubated with appropriate dilutions of specific primary antibodies against FAP (ab207178, 1:1000, Abcam) and α-SMA (ab124964, 1:1000, Abcam). The blots were incubated with HRP-conjugated secondary antibodies and visualized using the ECL system (GE, Boston, MA, USA).
Cell transfection
Indicated small interfering RNAs (siRNAs) were transfected into primary fibroblasts or cancer cells using Lipofectamine 3000 reagent (Invitrogen, Carlsbad, CA, USA). For stable transfection, the lentiviral vector (pGLV3/GFP/Puro) containing short hairpin RNAs (shRNAs) was designed based on the si-circNOX4#1 sequence and packaged by GenePharma (Shanghai, China). The circNOX4 sequence was packaged into the pGLV5/GFP/Puro vector. Fluorescent signals were observed using a fluorescence microscope (Olympus, Japan). The miRNA mimics, inhibitors and corresponding negative controls (NC) for miR-329-5p and miR-624-5p were synthesized and purchased from RiboBio (Guangzhou, China). The sequences used are listed in Additional file: Table
S5.
Cell viability assay
The cell viability of fibroblasts was measured by a cell counting kit-8 (CCK-8) assay (Solarbio, Beijing, China). A total of 2000 cells were seeded in 96-well plates per well, and 10 μl of CCK-8 reagent was added and incubated for 2 h at 37 °C. The absorbance was detected using a microplate reader (Tecan, Switzerland) at a wavelength of 450 nm.
Coculture of CAFs and cancer cells, conditioned medium (CM) preparation, migration and invasion assays
For coculture of CAFs and cancer cells, NSCLC cells were plated in the upper chamber of transwell apparatus (0.4 μm insert; Corning, Corning, NY, USA), and CAFs were cultured in the lower chamber. After incubation for 48 h, supernatants were collected for further measurements. For CM preparation, indicated fibroblasts were cultured to reach 80% confluence and then replaced with serum-free DMEM for 48 h. The supernatants were collected and centrifuged to remove cell pellets. The CM was subjected to cytological experiments or stored at -80 °C.
For the cell migration assay, NSCLC cells were cultured in six-well plates and scraped linearly to create an artificial wound. Then, NSCLC cells were incubated in CAF-CM or NF-CM for 48 h. For the cell invasion assay, the upper chambers of the Transwell plates (8 μm insert) were precoated with Matrigel (dilution, 1:8; BD Biosciences, Franklin Lakes, NJ, USA) at 37 °C for 30 min. NSCLC cells (5 × 10
4/well) suspended in serum-free medium were seeded into the upper chambers. Indicated fibroblasts (3 × 10
4/well) were seeded into the lower chambers. Subsequent procedures were performed according to previous studies [
20].
Dual luciferase reporter assay
Wild-type (WT) or mutant (MUT) pmiRGlo-circNOX4 and pmiRGlo-FAP-3′UTR dual luciferase reporter vectors incorporating predicted binding sites for miR-329-5p and miR-624-5p were constructed by GenePharma (Shanghai, China). The sequences are provided in Additional file: Table
S5 and Additional file: Table
S6. Cells were seeded on 12-well plates at a density of 60% and then co-transfected with WT or MUT plasmids with miRNA mimics or a negative control (NC) using Lipofectamine 3000 reagent. After 48 h of incubation, the cells were collected and tested for luciferase activity with a Dual-Luciferase Assay System kit (Promega, Madison, WI, USA) according to the manufacturer’s instructions.
RNA immunoprecipitation (RIP) assay
RIP was performed with an RNA immunoprecipitation kit (Geneseed, Guangzhou, China). CAFs were transfected with miR-NC or miR-329-5p for 48 h. Then, the cells were lysed with RIP lysis buffer and incubated with magnetic beads precoated with antibodies against Ago2 (ab186733, 1:50, Abcam) or IgG (ab172730, 1:50, Abcam). After the antibody was recovered by protein A/G beads and purification was completed, qRT-PCR was performed to detect the enrichment of circNOX4 and miR-329-5p in the precipitates.
Cytokine antibody array
A human cytokine array (AAH-CYT-G5; RayBiotech, Norcross, GA, USA) was used to measure the secretion levels of 80 cytokines in the CM of sh-circNOX4 CAFs and sh-NC CAFs according to the manufacturer’s instructions. Quantitative array analysis was performed using the ImageQuant LAS4000 Scanner (GE, Boston, MA, USA). Cytokines were screened using the following integrated conditions: Fold change ≥ 1.2 and fluorescence intensity values > 300.
Enzyme-linked immunosorbent assay (ELISA)
Cytokines in the supernatant of fibroblasts or cancer cells were measured using ELISA kits (IL-6, CCL2: Abcolonal, Wuhan, China; TGF-β2, IGFBP-4: RayBiotech, Norcross, GA, USA) following the manufacturer’s instructions.
Mouse experiments
The animal experiments were conducted in accordance with national guidelines and approved by the Animal Research Committee of the Fourth Hospital of Hebei Medical University (SYXK2022-011). BALB/c nude mice (female, age 5 weeks) were purchased from Beijing HFK Bio-Technology Co., Ltd.
For the subcutaneous xenograft model, 5 × 10
6 A549 cells alone or in a 3:1 mixture with stably transfected fibroblasts were subcutaneously implanted into the axillae of mice (
n = 6–8 mice per group). For the anti-IL-6 experiment group, neutralizing anti-IL-6 antibody (10 μg/ml, R&D, Minneapolis, MN, USA) was administered via intratumoral injection twice a week after 14 days of model construction [
21]. Tumor volume was calculated by using the formula: Volume = (length × width
2)/2. IHC analysis was performed on the xenografts to evaluate protein expression utilizing antibodies against FAP (ab207178, 1:250, Abcam), CD31 (28083–1-AP, 1:200, Proteintech), N-cadherin (Proteintech, 1:100, Servicebio), Vimentin (10366–1-AP, 1:400, Proteintech), matrix metalloproteinase 2 (MMP2) (ab86607, 1:200, Abcam), matrix metalloproteinase 9 (MMP9) (ab76003, 1:200, Abcam) and IL-6 (GB11117, 1:200, Servicebio).
For the lung metastasis model, 2 × 105 A549 cells in 100 μL PBS were injected into the tail vein of BALB/c nude mice. The mice were randomly assigned to one of three groups: (1) untreated A549 cells, (2) A549 cells "primed" in vitro for 72 h in a transwell coculture model with CAFs transduced with empty vector (sh-NC), or (3) CAFs transduced with circNOX4 shRNA (sh-circNOX4). After 5 weeks, the mice were sacrificed and dissected, and the lungs were collected. Hematoxylin and eosin (HE) staining was used to observe the metastatic foci.
68 Ga-labeled FAPI-04 PET imaging of mice
The preparation of
68 Ga-Labeled FAPI-04 solutions was performed by previous method [
22]. Briefly,
68 Ga was generated by a
68Ge–
68 Ga generator and eluted with a solution of 0.1 M hydrochloride. The mixture of the
68 Ga solution (74 MBq), 1.0 M sodium acetate (95 μL) and 1.15 mM FAPI-04 (20 μL) were reacted at 95 °C for 10 min.
68 Ga FAPI-04 PET imaging was performed up to 60 min after intravenous injection of 11.1 MBq of
68 Ga-FAPI-04 in tumor xenograft mice (
n = 3) using a micro-PET/SPECT/CT machine (Novel Medical Equipment Ltd., China) under isoflurane anesthesia. PET images were reconstructed with the comprehensive image analysis software Pmod v4.201 (PMOD Technologies LLC, Switzerland) and were converted to SUV images. Quantification was performed using a region-of-interest (ROI) technique and expressed as SUVmax.
Statistical analyses
Statistical analyses were performed using SPSS 22.0 software (SPSS Inc., Chicago, IL, USA), GraphPad Prism 7 (GraphPad, San Diego, CA, USA) and R software V4.2.0 (RStudio, Murray Hill, NJ, USA). The data are presented as the mean ± SD of at least three biological replicates. Differences between the two groups were analyzed by t-tests. The associations between circNOX4 expression and clinicopathological characteristics were evaluated by the Chi-square test. The statistical significance of survival was estimated by the Kaplan–Meier method and Cox analysis. P values less than 0.05 were considered statistically significant.
Discussion
Cancer metastasis is fostered by activated stroma, where specific microenvironments called niches are shaped [
1]. Cumulative studies have highlighted the important role of CAFs in establishing these niches that facilitate tumor growth and metastasis by secreting cytokines, chemokines, growth factors, and exosomes [
2]. However, how quiescent fibroblasts are educated into CAFs has not been extensively elucidated. In this study, we investigated the role and the underlying the mechanisms of FAP
+CAF activation and inflammatory niche formation. Based on clinical samples from NSCLC patients, we found that FAP
+CAFs correlate with distant metastasis and poor prognosis in NSCLC patients. Mechanistically, we determined that circNOX4 functions as a miR-329-5p sponge to upregulate FAP expression and induces NF to CAF phenotypic conversion. The activated CAFs subsequently enhances the production of IL-6. Since IL-6 has broad and robust effects on maintaining an inflammatory milieu and promoting cancer development, the in situ fibroblast niche boosts NSCLC metastasis and progression. Our findings highlight that targeting the circNOX4/miR-329-5p/FAP/IL-6 axis might be a novel strategy for cancer therapy in NSCLC.
In this study, we determined that FAP correlates with tumor metastasis and shortened survival of NSCLC patients. As a critical contributor to fibroblast activation, FAP is specifically expressed in the tumor stroma, which has prominent clinical implications [
27,
28]. For example, FAP-specific drugs have the potential to efficaciously target lesions in cancers [
29]. However, disappointing results were shown in the treatment of depleting FAP-positive cells in clinical trials of metastatic pancreatic and colorectal cancers [
30,
31]. Therefore, exploring the upstream modulators of fibroblast activation is of paramount importance.
Recently, the role of circRNAs in mediating the initiation and progression of cancers has been increasingly emphasized [
32]. Kristensen et al. demonstrated that circCDR1 was highly expressed in tumor stoma while its expression was absent in cancer cells, putting forward the notion that the intratumoral heterogeneity of circRNA expression should not be neglected [
33]. However, the roles and biological significance of circRNAs involved in CAFs remain largely unknown. We thus explored the potential involvement of circRNAs in primary fibroblasts from human NSCLC tissues. By conducting a circRNA array, we identified a novel CAF-specific circRNA, circNOX4. Clinically, the high expression of circNOX4 in the tumor stoma was correlated with distant metastasis and worse OS in NSCLC patients. Our study links circRNA expression in CAFs with the prognosis of NSCLC patients, suggesting the therapeutic potential of targeting circNOX4 in the TME. Moreover, we found that circNOX4 was responsible for the process of fibroblast activation. Gain- and loss-of-function experiments showed that knockdown of circNOX4 “normalized” CAFs to a quiescent phenotype and downregulated the expression of FAP, while overexpression of circNOX4 in NFs was sufficient to induce an activated fibroblast phenotype with higher FAP expression. Recent studies reported that miR-30a directly targeted FAP and suppressed its expression [
34]. Additionally, the transcription factor TCF21 reprogrammed FAP-high CAFs to a state that lacked pro-tumorigenic properties [
13]. Our study provides novel mechanisms by which circNOX4 increases the expression of FAP, the key marker of fibroblast activation, thereby contributing to the tumor-promoting phenotypes of CAFs.
The functions of circRNAs depend on their specific subcellular locations [
23]. Aiming to determine the molecular mechanisms of circNOX4, we detected its subcellular location in fibroblasts and discovered that it was mostly distributed in the cytoplasm. Cumulative studies have demonstrated that cytoplasmic circRNAs always sponge miRNAs and organize ceRNA regulatory networks to regulate gene expression [
23]. By cross-referencing the prediction results of the databases, miR-329-5p was identified as one of the candidate target miRNAs of circNOX4 and was possibly involved in CAF activation. We further confirmed the circNOX4/miR-329-5p/FAP axis by conducting qRT–PCR, FISH, luciferase reporter assays, RIP, and rescue experiments. In previous studies, miR-329-5p served as either an oncogene or a tumor suppressor in different cancers [
35‐
37]. A circRNA-mediated increase in miR-329-5p expression levels enhanced tumor proliferation and inhibited apoptosis in acute myeloid leukemia [
38]. However, the expression pattern, biological function and clinical significance of miR-329-5p in the tumor stroma have not been reported. Our study provides the first evidence that the circNOX4/miR-329-5p sponge in the ceRNA network is crucial for regulating FAP expression in CAFs, indicating a promising avenue for endogenously antagonizing FAP by inhibiting circNOX4.
CAFs support cancer cells through various paracrine pathways, including the secretion of inflammatory cytokines [
39]. As a versatile cytokine, IL-6 has broad and robust effects on tumor-promoting inflammation [
40], the EMT process [
41], and metastasis. IL-6 has also been recognized as a key marker of the inflammatory fibroblast subtype recently [
42]. In this study, we identified CAFs, rather than cancer cells, are the predominant source for secreting IL-6 in the local niche. We also determined that IL-6 as the downstream mediator of the circNOX4/FAP axis in CAFs, which contributed to NSCLC metastasis. This was further confirmed by the antitumor effects seen in vivo when blocking IL-6 with a neutralizing antibody, as multiple metastatic-related factors were suppressed. Likewise, researchers found that IL-6 produced by CAFs promotes chemoresistance through the JAK-STAT3 signaling pathway and that targeting IL-6 is a strategy to improve the therapeutic efficacy of chemotherapy in gastric cancer [
43]. It should be noted, however, that cytokines released by CAFs could circulate to distant organs and establish a premetastatic niche. Thus, it is possible that IL-6 fuels cancer cell dissemination from primary sites and colonization to distant organs, which requires further investigation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.