circWWC3 may promote triple-negative breast cancer by working with vimentin to control CSF2 release

The expression profiles of circRNAs in BC samples (n=5) and their corresponding adjacent non-tumorous tissues (n=5) were compared using RNA-Seq. These adjacent tissue samples were from the same patients as the tumors (paired samples), with tissues selected more than 2 cm away from the tumor margin serving as adjacent non-tumor tissue, and all samples were pathologically validated to ensure they were free of tumor cell infiltration. All RNA-Seq data comparisons were performed using a paired analysis method. RNA quality assessment and transcriptome analysis were performed by Guangzhou Geneseed Biotech Co., Ltd. Total RNA was extracted using TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc.), with quality verified by NanoDrop 2000 (A260/280>1.8, A260/230>2.0) and Agilent 2100 Bioanalyzer (RNA Nano 6000 Kit; RIN>7.0). Libraries were prepared using Total RNA-seq (H/M/R) Library Prep Kit for Illumina^® (Vazyme), quantified to 3.2 nM using Qubit 4.0 Fluorometer (dsDNA HS Assay Kit) and KAPA Library Quant Kit (Roche), then sequenced in 150 bp paired-end mode on Illumina NovaSeq 6000 platform. For bioinformatics analysis, the reads were mapped to the genome using STAR (https://github.com/alexdobin/STAR↗) and DCC (https://github.com/dieterich-lab/DCC↗) was used to identify circRNAs and estimate circRNA expression. TMM (trimmed mean of M-values) was used to normalize gene expression. Differentially expressed genes were identified using the edgeR program (v3.36.0) with a false discovery rate (FDR) threshold of <0.05 and |log2FC|>1. No covariates (such as age, sex, or disease stage) were included in the statistical model. For the analysis of cell mRNA microarrays, total RNA was extracted in triplicate from both the negative control MDA-MB-231 cells and the MDA-MB-231 cells with circWWC3 knockdown. The assessment of RNA quality and transcriptome analysis were conducted by Shanghai OE Biotech Co., Ltd. Differentially expressed genes were identified using thresholds of |log2FC|≥ 1 and p < 0.05.

In total, five paired fresh frozen tumor samples and their adjacent normal tissues were collected from patients with BC at the Breast Center of the Fourth Hospital of Hebei Medical University, with informed consent obtained from the patients. These patients had not received any treatment prior to surgery. The tissue microarray containing 150 TNBC tissues and 30 normal adjacent tissues (cat. no. HBreD180Bc01) were obtained from the Shanghai Xinchao Biological Technology Co. Ltd. The present study received approval from the Ethics Committee of the Fourth Hospital of Hebei Medical University (approval no. 2023KY045). All methods were performed in accordance with the relevant guidelines and regulations.

Normal human mammary epithelial cells MCF-10A and four types of human TNBC cells, namely MDA-MB-231, BT-549, MDA-MB-468 and HS-578T, were obtained from Procell Life Science & Technology Co., Ltd. MCF-10A cells were cultured in a specialized epithelial culture medium (Procell, China). MDA-MB-231 and HS-578T cells are cultured in DMEM, whilst MDA-MB-468 cells were cultured in Leibovitz's L-15 medium (GIBCO; Thermo Fisher Scientific, Inc.) and BT-549 cells were grown in RPMI 1640 medium (GIBCO; Thermo Fisher Scientific, Inc.). All four triple-negative breast cancer cell lines are cultured with 10% FBS and supplemented with 1% penicillin and streptomycin. Additionally, HS-578T and BT-549 cells are supplemented with 10 μg/ml insulin. All cell lines were maintained at 37˚C in a humidified incubator with 5% CO2. For transfection, circWWC3 small interfering (si)RNA (sense: 5'-GAAAGAGGATAACAAAGCC-3'; antisense: 5'- GGCUUUAUUACUCUUUC-3'), the si-NC (sense: 5'-CGUACGCGGAAUACUUCGA-3'; antisense: 5'-UCGAAGUAUUCCGCGUACG-3'), and vimentin overexpression (pCMV-VIM(human)-EGFP-Neo) constructs were obtained from Haixing Biosciences. Overexpression constructs for circWWC3 and the control (pLC5-ciR) vector were sourced from Guangzhou Geneseed Biotech Co., Ltd. A total of 3 μg plasmid DNA was transfected using FuGENE HD (Promega Corporation) for the overexpression experiments. For siRNA transfection, MDA-MB-231 and BT-549 cells were transfected with HiPerFect (Qiagen GmbH) according to the manufacturer's protocols.

Total RNA was extracted from MDA-MB-231 and BT-549 cells using TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc.). In the actinomycin D assay, MDA-MB-231 and BT-549 cells were seeded in 6-well plates prior to treatment with actinomycin D (AdooQ Bioscience, USA). All cells were exposed to 2 μg/mL actinomycin D for durations of 4, 8, 12, and 24 hours, while the control group remained untreated. For the RNase R assay, 2 μg RNA was treated with 3 U/μg RNase R (Epicentre; Illumina, Inc.) at 37°C for 15 min, followed by a 3-min incubation at 85°C. A parallel control sample was processed identically but in the absence of RNase R. Following the RNase R and actinomycin D treatments, the expression levels of circWWC3 and linear WWC3 were assessed using reverse transcription-quantitative reverse transcription PCR (RT-qPCR).

Total RNA was extracted from the cells using TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc.). cDNA was synthesized using GoScript™ (Promega Corporation). qRT-PCR was conducted using SYBR Green Reagent (Promega Corporation). The primer sequences used for PCR amplification are provided in Table 1. U6 snRNA and GAPDH served as internal controls for circRNAs and mRNAs, respectively.

Cytoplasmic and nuclear RNA isolation was performed using the Thermo Scientific NE-PER nuclear and cytoplasmic extraction kit (Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. Briefly, cells were treated with cell fractionation buffer and incubated on ice for 5 min before centrifugation. This process facilitated the separation of the cytoplasmic fraction from the nuclear pellet. Subsequently, the samples were incubated at 25°C with a lysis/binding solution to differentiate the nuclear and cytoplasmic RNA. The RNA for both fractions was then isolated using a filter cartridge and eluted with an elution solution. The isolated samples were analyzed using RT-qPCR.

The Cy3-labeled circWWC3 probe was purchased from Geneseed Biotech Co., Ltd. (Guangzhou, China), and the FISH assay was performed using a commercial kit (Geneseed Biotech) following the manufacturer's protocol. For cell-based FISH, MDA-MB-231 and BT-549 cells were grown on coverslips, fixed in 4% formaldehyde (20 min, RT), and permeabilized with 0.2% Triton X-100 in PBS. For tissue FISH, TNBC tissue microarray slides were treated to remove paraffin using xylene and ethanol solutions, followed by a 10-min incubation with proteinase K at 37°C for blocking. The FISH probes were diluted to 20 nM, denatured, and pre-hybridized before application to cells or tissues. Hybridization was carried out overnight at 37°C. After hybridization, nuclei were counterstained with DAPI, and images were captured using a confocal laser scanning microscope (Carl Zeiss AG). The sequences of the probes are as follows: circWWC3-Cy3 5'Cy3-TCAATGGCTTTGTTATCCTCTTTCT-3'Cy3.

Cell proliferation was assessed using the Cell Counting Kit-8 (CCK-8) assay. MDA-MB-231, BT-549, MDA-MB-468 and HS-578T (2x10³ cells per well) cells were seeded into a 96-well plate, before 10 μl CCK-8 solution was added to each well daily. After 1 h incubation at 37°C, the absorbance at 450 nm was measured using a multifunctional microplate reader. For colony formation assays, MDA-MB-231, BT-549, MDA-MB-468 and HS-578T cells were stored in 6-well plates at the density of 2,000 cells per well. The cells were then placed in a 37°C sterile incubator for 12 days. Afterwards, the cell colonies were stained with crystal violet solution and subsequently observed.

Cell migration and invasion abilities were assessed using a Transwell chamber (Corning, Inc.). For invasion assays, Matrigel (BD Biosciences) was evenly spread in the chamber at 37°C overnight. Migration assays were performed using Transwell chambers without Matrigel. Cells (4x10⁴) were placed into the upper chambers containing serum-free culture medium, whilst the lower chamber was filled with culture medium supplemented with 10% FBS. After 24–48 h, the cells in the upper chamber were wiped off and the MDA-MB-231, BT-549, MDA-MB-468 and HS-578T cells that had invaded to the bottom of the membrane were stained with crystal violet. Cells were counted in five randomly selected fields per membrane using ImageJ software (v1.53; National Institutes of Health). The mean value from three independent experiments was used for statistical analysis.

The protein expression levels of CSF2 in cell supernatants were quantified using a CSF2 Human ELISA Kit (ProteinTech Group, Inc.) following the manufacturer's protocols.

The RNA pull-down probe targeting the junction site of circWWC3 was synthesized by Guangzhou RiboBio Co., Ltd. MDA-MB-231 and BT-549 cells were first lysed using a lysis buffer and subsequently incubated with the biotin-labeled circWWC3 probe (circWWC3 probe/5'BiosG/TCAATGGCTTTGTTATCCTCTTTCT/3'Bio/). The cell lysates were then combined with streptavidin-coated agarose magnetic beads to capture the biotin-labeled RNA-protein complexs, utilizing the Pierce™ Magnetic RNA-Protein Pull-Down Kit (Thermo Fisher Scientific, Inc.). After incubation, the beads were collected using a magnetic stand, washed three times with Wash Buffer to remove nonspecifically bound proteins, and the RNA-binding proteins were eluted by incubation with Elution Buffer (containing competitive ligands or denaturing agents) at room temperature for 15 min with gentle agitation. The proteins isolated by probes were identified via Western blot analysis. Further mass spectrometry analysis was performed using QLBio (Beijing, China). Proteins were identified and quantified using the Proteome Discoverer Software 2.4 (Thermo Fisher Scientific). The mass spectrometry data were searched against the UniProtKB Human database using the Sequest HT search engine. The search parameters were set as follows: trypsin as the enzyme with up to two missed cleavages allowed; precursor mass tolerance of 15 ppm and fragment mass tolerance of 0.02 Da; carbamidomethylation (C) was set as a fixed modification; oxidation (M) and acetylation (protein N-terminus) were set as variable modifications. To ensure high-confidence identifications, the false discovery rate (FDR) was controlled at ≤ 1% at both the peptide-spectrum match (PSM) and protein levels using the Percolator node. Protein identifications were required to contain at least one unique peptide.

MDA-MB-231 and BT-549 cells were seeded onto glass cover slips and fixed at room temperature using 4% paraformaldehyde, followed by a 1-h rehybridization step. The cells were then treated overnight at 37°C with a Cy3-labeled circWWC3 probe. After blocking with 10% BSA (cat. no. A8010; Solarbio, Beijing, China) at 37°C for 30 mins, the cells were incubated for 1 h at room temperature with a vimentin antibody (cat. no. 10366-1-AP; ProteinTech Group, Inc.), followed by a 1-h incubation at 37°C with an Alexa Fluor™ 488-conjugated secondary antibody (cat. no. SA00013-2; ProteinTech Group, Inc.). Finally, the coverslips were sealed with parafilm-containing DAPI (cat. no. SI103-11; Seven, Beijing, China) and images were captured using a confocal microscope.

MDA-MB-231, BT-549, MDA-MB-468 and HS-578T cells from each group were lysed using RIPA lysis buffer (cat. no. SW104-2; Seven, Beijing, China) supplemented with PMSF (1 mM). The resulting protein lysates were denatured through heat treatment and subjected to electrophoresis on 10% SDS polyacrylamide gels. After separation, the proteins were transferred onto PVDF membranes and blocked with 5% skimmed milk. Primary antibodies against E-cadherin (cat. no. 20874-1-AP; 1:1,000; ProteinTech Group, Inc.), N-cadherin (cat. no. 22018-1-AP; 1:1,000; ProteinTech Group, Inc.), Vimentin (cat. no. 10366-1-AP; 1:1,000; ProteinTech Group, Inc.), phosphorylated (p-)-Vimentin (S56; cat. no. ab217673; 1:1,000; Abcam) and β-actin (cat. no. 20536-1-AP; 1:1,000; ProteinTech Group, Inc.) were added and incubated overnight at 4°C. Following this, secondary antibodies (cat. no. SA00001-1/SA00001-2; 1:5,000; ProteinTech Group, Inc.) were applied for 1 h. The membranes were developed using an ECL kit (Beijing Solarbio Science & Technology Co., Ltd.). The intensity of the immunoblot lanes was quantified using the Image J software (v1.53; National Institutes of Health).

RIP assays were conducted using a commercial RIP Kit (Geneseed, Guangzhou, China) according to the manufacturer's protocol. In brief, MDA-MB-231 cells transfected with Flag-Vimentin-WT or Flag-Vimentin-S56A plasmids for 48 hours were collected (2×10⁷ cells), lysed, and centrifuged. The resulting supernatants were incubated overnight at 4°C with magnetic beads conjugated with either anti-Flag antibody or control IgG (5 μg per sample; Proteintech, China). Subsequently, RNA was isolated from the immunoprecipitated complexes, and relative expression levels of target RNAs were measured via RT-qPCR.

The cytotoxic activity of NK-92MI cells toward TNBC cells was evaluated by measuring LDH release with the CytoTox 96^® Non-Radioactive Cytotoxicity Assay Kit (cat. no. G1780, Promega, Madison, USA). Briefly, breast cancer cells were transfected with the following constructs: si-NC, si-circWWC3, si-NC + vector, si-circWWC3 + vector, and si-circWWC3 + vimentin. After 24 hours, NK-92MI cells were co-cultured with the transfected TNBC cells at effector-to-target ratios of 5:1, 10:1, and 20:1 for 6 hours. Subsequently, the culture supernatants were collected and analyzed for LDH activity.

The in vivo experiments received approval from the Institutional Animal Care and Use Committee of the Fourth Affiliated Hospital of Hebei Medical University (approval no. IACUC-4th Hos Hebmu-2023045). All BALB/c nude mice were sourced from Hfk Bioscience Co., Ltd. (Beijing, China). In the xenograft tumor study, 4-week-old female BALB/c nude mice were randomly assigned (n=5 per group) to one of three experimental groups: i) MDA-MB-231 cells stably transfected with shControl + Vector; ii) shcircWWC3 + Vector; or iii) shcircWWC3 + Vimentin, with the experimenter blinded to the treatment regimen during all measurements. The cells were then injected subcutaneously into the right mammary fat pad of each mouse (5x10⁶ cells in 100 μl PBS) to establish the xenograft model. To ensure ethical compliance, tumors were allowed to grow until they reached a maximum volume of 1.5 cm³. After tumor formation, the volume was measured every 3 days to calculate tumor growth. At the end of the experiment, the mice received an intraperitoneal injection of ketamine (100 mg/kg) in combination with xylazine (10 mg/kg), both prepared in physiological saline. The mice were then humanely euthanized by cervical dislocation, and tumors were harvested and weighed. All the experiments were conducted following the ARRIVE guidelines.

The structural configurations of circWWC3 and vimentin protein were computationally predicted using RNACOMPOSER (v1.0; http://rnacomposer.ibch.poznan.pl/↗) and AlphaFold2 (v2.3.2; https://www.alphafold.ebi.ac.uk/↗), respectively, followed by structural optimization in AutoDock Vina (v1.1.2; http://vina.scripps.edu/↗) through protonation state adjustment, atomic completeness restoration, and motif reconstruction. Molecular docking analysis was then performed using HDOCK (v1.1; http://hdock.phys.hust.edu.cn/↗) with the protein structure maintained as rigid body under comprehensive surface scanning conditions, generating 100 docking conformations from which the most thermodynamically favorable complex (determined by minimal binding energy score) was selected for three-dimensional visualization and analysis using PyMOL (v2.4.0; https://pymol.org/2/↗).

Immunohistochemistry (IHC) was employed to evaluate protein expression in tumor tissue specimens. Following deparaffinization of tumor sections, endogenous peroxidase activity was blocked with 3% hydrogen peroxide. The sections were then incubated overnight at 4°C with primary antibodies targeting CSF2 (cat. no. 17762-1-AP; 1:500; ProteinTech Group, Inc.), after which HRP-conjugated secondary antibodies were applied. Antigen visualization was achieved using diaminobenzidine (DAB), and hematoxylin was used for nuclear counterstaining. IHC assessment was conducted in a blinded manner, following the same scoring methodology applied in the FISH assay. Specimens were categorized as negative (0), weak (1–2), moderate (3), or strong (4–6) according to the composite scoring system.

Statistical analyses for all experiments were conducted using SPSS 24.0 (IBM Corp) and GraphPad Prism software (Dotmatics). Data are expressed as mean ± SD. Comparisons between groups were conducted using paired Student's t-test or one-way analysis of variance (ANOVA) followed by Dunnett's test or Tukey's test. The χ2 test was used to assess the relationship between gene expression and the various clinicopathological characteristics. P<0.05 was considered to indicate a statistically significant difference.

CircRNA expression patterns were examined in five paired samples of BC using the Illumina sequencing platform. Relative to normal tissues, the BC tissue group exhibited a notable upregulation of 224 circRNAs and a marked downregulation of 231 circRNAs (Figure 1A). Hsa_circ_0001910, derived from WWC3 and designated as circWWC3, was found to be significantly upregulated among the differentially expressed circRNAs. Therefore, circWWC3 was chosen as the primary focus of the present study.

circWWC3 originates from exons 2–8 of the host gene WWC3 (Figure 1B). Initially, the expression of circWWC3 was assessed in both normal breast epithelial cells (MCF-10A) and TNBC cells (MDA-MB-231, BT-549, MDA-MB-468 and HS-578T). MDA-MB-231 and BT-549 cells, which are TNBC cell lines, consistently demonstrated elevated expression of circWWC3 (Figure 1C). Based on the results of the actinomycin D and RNase R assays, circWWC3 exhibited greater stability compared with WWC3 mRNA following treatment with actinomycin D (Figure 1D) and RNase R (Figure 1E). Subsequent nuclear-cytoplasmic fractionation and FISH assays demonstrated that circWWC3 was predominantly localized in the cytoplasm instead of the nucleus of MDA-MB-231 and BT-549 cells (Figures 1F, G). These aforementioned findings indicated a significant increase in circWWC3 expression levels in both TNBC tissues and cells, where the primary localization was the cytoplasm in TNBC cells.

To assess the expression level of circWWC3 in a larger population, a tissue microarray that included 150 TNBC tissues along with 30 corresponding adjacent tissues was obtained. It was found that the expression of circWWC3 was significantly increased in TNBC tissues compared with that in the corresponding adjacent tissues (Figures 2A, B). The association between circWWC3 expression levels and the clinicopathological characteristics of patients with TNBC was next examined. It was observed that patients with advanced T stage and lymph node metastasis demonstrated elevated levels of circWWC3 expression (Figures 2C, D) (Table 2). Analysis of the Kaplan-Meier plotter database also revealed that the expression of WWC3, the parental gene of circWWC3, is associated with poor patient prognosis (Supplementary Figure S2). These aforementioned results suggest that circWWC3 is highly expressed in TNBC, where its elevated expression is associated with the patient's T stage and lymph node metastasis.

To further investigate the impact of circWWC3 on the biological functions of TNBC, gain-of-function and loss-of-function experiments were conducted. RNA interference was used to suppress circWWC3 expression and confirmed the knockdown efficiency in MDA-MB-231 and BT-549 cells (Supplementary Figure S1A). The CCK-8 and clone formation assay results indicate that circWWC3 knockdown significantly inhibited the proliferation and clone formation capabilities of MDA-MB-231 and BT-549 cells (Figures 2E, G). Transwell cell invasion and migration assays found that circWWC3 knockdown significantly inhibited the invasion and migration of MDA-MB-231 and BT-549 cells (Figure 2I).

The circWWC3 overexpression plasmid was then constructed, which were transfected into MDA-MB-468 and HS-578T cells. RT-qPCR analysis demonstrated that the expression level of circWWC3 was markedly increased upon circWWC3 overexpression (Supplementary Figure S1B). CCK8 assay and colony formation assay indicated that the overexpression of circWWC3 effectively enhanced the proliferation and clone formation capabilities of MDA-MB-468 and HS-578T cells (Figures 2F, H). The promoting effects of cell migration and invasion were also demonstrated using cell migration and invasion assays (Figure 2J). The aforementioned data collectively indicate that circWWC3 may contribute to the development and progression of TNBC.

To explore the downstream signaling pathways associated with circWWC3, an mRNA expression profile microarray analysis was next conducted in MDA-MB-231 cells transfected with si-negative control (NC) or si-circWWC3. These results indicate that after si-circWWC3 transfection, a total of 682 genes exhibited changes compared with the si-NC group, with 379 genes being upregulated and 303 genes downregulated (Figure 3A). The Kyoto Encyclopedia of Genes Genomes pathway enrichment analysis revealed that circWWC3 is involved in the regulation of cytokine-cytokine receptor interactions, tight junctions and natural killer cell-mediated cytotoxicity, among others (Figure 3B). A selection of representative genes was chosen and validated through RT-qPCR, with CSF2 being one of the genes that exhibited the most significant changes (Figure 3C). Several studies have previously identified CSF2 as a chemotactic factor for tumor cells, demonstrating its strong association with proliferation, invasion and metastasis (21, 22). Subsequent overexpression or knockdown experiments of circWWC3 indicated that overexpressing circWWC3 led to an increase in CSF2 levels (Figure 3D), whilst knocking down circWWC3 resulted in a decrease in CSF2 levels (Figure 3E), according to ELISA performed on the culture supernatants of TNBC cells. Notably, circWWC3 overexpression had minimal impact on CSF2 mRNA levels but significantly enhanced CSF2 protein secretion, suggesting that the regulation likely occurs at the post-translational or secretory level rather than at the transcriptional level.

As the KEGG pathway analysis indicated that circWWC3 is involved in NK cell-mediated cytotoxicity, we performed NK cell cytotoxicity assays. The results demonstrated that the cytotoxicity of NK-92MI cells was significantly enhanced when co-cultured with circWWC3-knockdown MDA-MB-231 and BT-549 cells compared to the si-NC group (Figure 3F). To verify whether the biological function of circWWC3 in TNBC cells is mediated by the stimulation of CSF2, rescue experiments involving circWWC3 and CSF2 were conducted. The results indicated that knockdown of CSF2 partially reversed the effects of circWWC3 overexpression on the proliferation (Figure 4A) and clonogenic ability (Figure 4B) of MDA-MB-468 and HS-578T cells. In addition, the results of the Transwell cell invasion and migration experiments indicated that knocking down CSF2 partially reversed the effects of circWWC3 overexpression on the invasion and migration abilities of MDA-MB-468 and HS-578T cells (Figures 4C, D). In summary, these results indicate that the overexpression of circWWC3 promotes the malignant progression of TNBC cells by enhancing the secretion of CSF2.

circRNAs has been previously reported to serve a role in molecular regulation through their interactions with proteins (10). To investigate the proteins that can associate with circWWC3, RNA pull-down and mass spectrometry assay were conducted to identify its interacting proteins. An interaction between the vimentin protein and circWWC3 was observed (Figure 5A; Table 3), which was validated using western blot analysis (Figure 5B). Western blotting results indicated that knocking down circWWC3 markedly reduced the expression of vimentin and N-cadherin in MDA-MB-231 and BT-549 cells, whilst markedly increasing that of E-cadherin (Figure 5C). Molecular-docking analysis indicated that circWWC3 is folded to form a DNA-like double helix structure, which enhances the stability of the RNA. The folded circWWC3 encases the vimentin protein and aligns closely with it, facilitating the formation of a stable RNA-protein complex (Figure 5D). RNA FISH immunofluorescence analysis confirmed the co-localization of circWWC3 and vimentin in the cytoplasm of MDA-MB-231 and BT-549 cells (Figure 5E). Furthermore, the target sites for circWWC3 on vimentin were confirmed through RNA pulldown and western blotting using antibodies against p-vimentin (S56) (Figure 5F). RNA immunoprecipitation (RIP) assays validate that the binding of circWWC3 to vimentin is dependent on the Ser56 residue (Figure 5G). These results suggest that circWWC3 can bind to vimentin at the S56 site. A previous study has reported that circRNAs can bind to vimentin and influence the EMT in tumor cells upstream of tumor progression (23). Additionally, other studies have indicated that the EMT in breast cancer cells can enhance the secretion of CSF2 (24). Therefore, it was next hypothesized that this interaction may serve a role in the secretion of CSF2. Notably, the downregulation of CSF2 induced by circWWC3 knockdown was partially reversed upon vimentin overexpression. However, this reversal effect was abolished when vimentin was mutated to Vimentin (S56A) (Figure 5H). The NK-92MI cell cytotoxicity results demonstrated that the enhanced killing ability induced by circWWC3 knockdown could be counteracted by vimentin overexpression (Figure 5I). Furthermore, vimentin overexpression partially recovered the abilities of proliferation (Figure 6A) and colony formation (Figure 6B), which were originally inhibited by circWWC3 knockdown. Similarly, the suppressive effects on migration and invasion caused by the knockdown of circWWC3 were partially reversed through the overexpression of vimentin (Figure 6C). Moreover, the inhibition of EMT in TNBC induced by circWWC3 knockdown was found to be reversed by vimentin overexpression (Figure 6D).

To explore the relationship between circWWC3 and the growth of TNBC tumors, MDA-MB-231 cells stably transfected with shControl + vector, shcircWWC3 + vector or a combination of shcircWWC3 + vimentin vectors were subcutaneously injected into BALB/c nude mice. Knocking down circWWC3 significantly inhibited the growth of xenograft tumors. However, this inhibitory effect was reversed when vimentin was overexpressed (Figures 7A–D). We also examined the expression of CSF2 in the aforementioned groups using IHC. Our results indicated that CSF2 expression was markedly reduced upon circWWC3 knockdown, and this inhibitory effect was counteracted when vimentin was overexpressed simultaneously with circWWC3 knockdown (Figure 7E). Overall, these findings suggest that vimentin-mediated activation of the EMT signaling pathway may significantly influence the promoting effects of circWWC3 on the progression of TNBC, both in vitro and in vivo.