A non-coding RNA circGNAI2 suppresses triple-negative breast cancer progression by sponging miR-454-3p to inhibit the VGLL4-STAT3 signaling

Abstract

Circular RNAs (circRNAs), a novel group of non-coding RNAs, have been involved in the occurrence and development of different malignancies. However, the potential functional role of circRNAs in triple-negative breast cancer (TNBC) remains largely unknown, we aimed to investigate the regulatory functions and underlying mechanism of a conserved circRNA, circGNAI2, in TNBC. In this study, we demonstrated that circGNAI2 was aberrantly downregulated in TNBC cells and was negatively associated with poor prognosis. Functional experiments showed that circGNAI2 significantly suppressed TNBC cell proliferation and migration in vivo and in vitro. Mechanistically, we found that circGNAI2 upregulated VGLL4 and inhibited the phosphorylation STAT3 by sponging miR-454-3p. Moreover, EIF4A3 has binding sites in the upstream region of the GNAI2 pre-mRNA, and we confirmed that circGNAI2 was downregulated by EIF4A3 in TNBC cells. Overall, our study indicates that EIF4A3-regulated circGNAI2 suppresses TNBC progression possibly by regulating the miR-454-3p/VGLL4/STAT3 pathway, providing pivotal potential therapeutic targets for the treatment of TNBC.

Breast cancer (BC) is the most common malignancy among women worldwide (1, 2). Triple-negative breast cancer (TNBC) is a distinctive and highly aggressive subtype of BC characterized by the absence of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER-2). Consequently, TNBC is insensitive to both endocrine therapy and HER2-targeted treatment modalities (3) and shows a high rate of recurrence, metastasis, and poor prognosis compared with other molecular types of BC (4, 5). Hence, investigation of the potential mechanisms underlying TNBC tumorigenesis and identification of novel, effective therapeutic targets for TNBC have emerged as imperative clinical requirements.

Circular RNAs (circRNAs) are a group of stable non-coding RNAs with a unique circular structure generated by back-splicing (6). Currently, an extensive study of circRNAs is underway in the field of cancer research, with numerous studies indicating that circRNAs are involved in the processes of progression, recurrence, and metastasis (7, 8). Many circRNAs have been identified as oncogenes or tumor suppressors involved in the malignant progression of BC (9). For example, circPRKCI promotes WBP2 expression by sponging and regulating miR-545-3p (10). CircSEMA4B synergizes with SEMA4B-211aa, a novel peptide encoded by CircSEMA4B, to suppress aberrant activation of the PI3K/AKT signaling pathway, thereby inhibiting BC progression (11). Moreover, because of their unconventional circular structures, circRNAs exhibit higher stability than linear RNAs, resisting exonuclease degradation and thereby playing exclusive roles in cellular regulation (12). Therefore, circRNAs have the potential to serve as robust biomarkers, efficacious therapeutic targets, and interventions for tumor treatment (13).

By analyzing breast cancer-related circRNA microarray dataset GSE182471 (from GEO), we found that circGNAI2, alternately referred to as hsa_circ_0007083, was downregulated in majority of breast cancer tissues (4/5) compared to adjacent normal tissues (Fig. S1A). Furthermore, the microarray dataset GSE101124 (from GEO) revealed that lower circGNAI2 expression in TNBC than in luminal A subtype (Fig. S1A). Based on these findings, we identified circGNAI2 as a candidate for further study. To validate our initial screening results, we first examined circGNAI2 expression in TNBC samples and cell lines. Here, we found that circGNAI2 was downregulated in TNBC tissues and cell lines. CircGNAI2 exerted tumor-suppressive effects in TNBC, possibly via regulating the miR-454-3p/VGLL4/STAT3 pathway. Furthermore, we identified EIF4A3 as a regulator of circGNAI2 expression. Our findings revealed a novel regulatory mechanism in TNBC, providing a new avenue for TNBC therapy.

Results

The expression of circGNAI2 was downregulated in TNBC tissues and cell lines

According to the annotation of the UCSC Genome Browser (http://genome.ucsc.edu/), circGNAI2 is formed by the back-splicing of exon 3 to 5 of gene GNAI2. Sanger sequencing verified the back-splicing junction of circGNAI2 (Fig. 1A). PCR assays indicated that circGNAI2 could be produced by divergent primers with cDNA rather than genomic DNA (gDNA), whereas the linear isoform of GNAI2 could be produced by convergent primers with both cDNA and gDNA (Fig. 1B). We subsequently examined circGNAI2 expression in 46 randomly selected TNBC tissues and paired adjacent normal tissues and found that circGNAI2 expression was significantly lower in TNBC tissues than in adjacent normal tissues (Fig. 1C). We also analyzed the correlation between circGNAI2 expression and clinicopathological variables in 46 patients. The results showed that circGNAI2 expression was negatively associated with lymph node metastasis and TNM tumor stage (Table 1). Similarly, the results showed that cirGNAI2 expression was significantly lower in TNBC cells (MDA-MB-231, BT-549, HCC-1937, MDA-MB-468, HS578T) and HER2-amplified cell lines (SK-BR-3) than that in MCF-10A cells, meanwhile, we found that cirGNAI2 expression in the Luminal A cell lines (MCF7, T47D) is relatively higher than that in TNBC cells (MDA-MB-231, BT-549, HCC-1937, MDA-MB-468, HS578T) (Fig. 1D). Subsequently, we verified the higher stability and longer half-life of circGNAI2 compared to linear GNAI2 mRNA using RNase R treatment (Fig. 1, E and F) and Actinomycin D assays (Fig. 1, G and H). The subcellular localization of circular RNAs (circRNAs) is closely associated with their functional mechanisms. Generally, cytoplasmic circRNAs predominantly regulate cellular pathways by acting as microRNA (miRNA) sponges or by binding to RNA-binding proteins (RBPs), whereas nuclear circRNAs primarily modulate gene expression and transcriptional processes (14, 15). We then performed a subcellular fractionation assay and found that circGNAI2 was predominantly located in the cytoplasm of MDA-MB-231 and HCC-1937 cells (Fig. 1, I and J). Consistently, the FISH assay also suggested that circGNAI2 was mainly stained in the cytoplasm of the MDA-MB-231 and HCC-1937 cells (Fig. 1K).

Figure 1.

The expression of circGNAI2 was downregulated in TNBC tissues and cell lines.A, circGNAI2 is formed by circularization of exon 3 to 5 of the gene GNAI2, and the splicing junction was verified by Sanger sequencing. B, existence of circGNAI2 in MDA-MB-231 and HCC-1937 cell lines was verified by agarose gel electrophoresis. C, circGNAI2 had low expression in TNBC tissues compared with adjacent normal tissues. D, relative expression of circGNAI2 in BC cell lines. E-F, RT-qPCR analysis of circGNAI2 and linear GNAI2 in MDA-MB-231 and HCC-1937 cell lines treated with RNase R. G-H, after Actinomycin D treatment, the mRNA stability of circGNAI2 and GNAI2 in MDA-MB-231 and HCC-1937 cell lines were determined by RT-qPCR. I-J, expression levels of cytoplasmic control transcripts (GAPDH), the nuclear control transcript (U6), and circGNAI2 were determined by RT-qPCR in the cytoplasmic and nuclear fractions of MDA-MB-231 and HCC-1937 cell lines. K, FISH experiments were conducted to ascertain the subcellular localization of circGNAI2 in MDA-MB-231 and HCC-1937 cells. Red, circGNAI2; Blue, DAPI. ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Table 1.

Relationship between circGNAI2 expression and clinicopathologic characteristics in TNBC

Characteristics	No. of cases	circGNAI2 expression		p value
		≥0.701	<0.701
Age (year)				0.2033
≥50	25	13	12
<50	21	7	14
Tumor size (mm)				0.0597
≥20	32	11	21
<20	14	9	5
Ki-67 index				0.4299
≥30	41	17	24
<30	5	3	2
Lymph node metastasis				0.0273
No	31	10	21
Yes	15	10	5
Tumor recurrence or metastasis				0.9713
No	39	17	22
Yes	7	3	4
TNM tumor stage
I + II	34	11	23	0.0104
III + IV	12	9	3

CircGNAI2 suppresses the proliferation and migration abilities of TNBC cells

To verify the biological function of circGNAI2 in TNBC cells, we transfected MDA-MB-231 and HCC-1937 cells with two specific siRNAs targeting circGNAI2 (si-circGNAI2-1 and si-circGNAI2-2) (Fig. 2A). Subsequent MTT and colony formation assays demonstrated that silencing circGNAI2 promoted the proliferative ability of MDA-MB-231 and HCC-1937 cells (Fig. 2, B–E). Furthermore, transwell and wound-healing assays confirmed that silencing circGNAI2 promoted the migration ability of MDA-MB-231 and HCC-1937 cell lines (Fig. 2, F– I). We then constructed a circGNAI2 plasmid (circGNAI2), which stably overexpressed circGNAI2 compared to the control (NC) (Fig. 3A). Upon increasing the expression level of circGNAI2, the proliferation (Fig. 3, B–E) and migration abilities of MDA-MB-231 and HCC-1937 cells were impaired (Fig. 3, F–I). Taken together, circGNAI2 inhibits the proliferation and migration of TNBC cells in vitro.

Figure 2.

Silencing circGNAI2 promotes the proliferation and migration ability of TNBC cells.A, expression of circGNAI2 was confirmed by RT-qPCR in MDA-MB-231 and HCC-1937 cell lines transfected with si-NC or si-circGNAI2. B-C, effect of si-circGNAI2 on proliferation in MDA-MB-231 and HCC-1937 cell lines by MTT assay. D-E, effect of si-circGNAI2 on proliferation in MDA-MB-231 and HCC-1937 cell lines by colony formation assay. F-G, effect of si-circGNAI2 on migration in MDA-MB-231 and HCC-1937 cell lines by transwell assay. H-I, effect of si-circGNAI2 on migration in MDA-MB-231 and HCC-1937 cell lines by wound healing assay. ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Figure 3.

Overexpression of circGNAI2 suppresses the proliferation and migration abilities of TNBC cells.A, expression of circGNAI2 was confirmed by RT-qPCR in MDA-MB-231 and HCC-1937 cell lines transfected with NC or circGNAI2. B-C, effect of circGNAI2 on proliferation in MDA-MB-231 and HCC-1937 cell lines by MTT assay. D-E, effect of circGNAI2 on proliferation in MDA-MB-231 and HCC-1937 cell lines by colony formation assay. F-G, effect of circGNAI2 on migration in MDA-MB-231 and HCC-1937 cell lines by transwell assay. H-I, effect of circGNAI2 on migration in MDA-MB-231 and HCC-1937 cell lines by wound healing assay. ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

CircGNAI2 serves as a sponge for miR-454-3p

As circGNAI2 is mainly expressed in the cytoplasm of TNBC cells, we first explored the possibility of its functioning as a miRNA sponge. The prediction results from three online databases, circBank, miRanda, and starBase, were analyzed. We identified six miRNAs that were most likely to bind to circGNAI2: miR-19a-3p, miR-135a-5p, miR-135b-5p, miR-301b-3p, miR-330-3p, and miR-454-3p (Fig. 4A). We examined the changes in the expression levels of the six miRNAs after circGNAI2 overexpression in MDA-MB-231 and HCC-1937 cells, and found that only miR-454-3p was significantly downregulated in both cell lines (Fig. 4, B and C). Subsequent circGNAI2 knockdown experiments produced the inverse effect, with miR-454-3p displaying specific upregulation among all miRNAs examined in MDA-MB-231 and HCC-1937 cell lines (Fig. S1B). The putative binding sites between circGNAI2 and miR-454-3p were predicted using the online database. Plasmids containing WT or MUT sequences of putative binding sites were constructed (Fig. 4D). In vivo, mature microRNAs (miRNAs) form RNA-induced silencing complexes (RISC) with Argonaute 2 (AGO2), mediating targeted mRNA degradation, or translational repression (16, 17). CircRNAs can function as competitive endogenous RNAs (ceRNAs) by sponging miRNAs, thereby attenuating miRNA-mediated gene silencing. The interaction between miRNAs and circRNAs can be indirectly validated through AGO2 RNA immunoprecipitation (RIP), AGO2-RIP assay demonstrated significant enrichment of circGNAI2 in AGO2 immunoprecipitates, suggesting that circGNAI2 interacts with miRNAs in an AGO2-dependent manner (Fig. 4E). A luciferase reporter assay was performed to further prove that, in comparison to the NC, miR-454-3p-mimics obviously decreased the relative activity (Firefly/Renilla) of the WT group. However, in the MUT group, no significant difference in relative luciferase activity (Firefly/Renilla) was observed between the NC and miR-454-3p mimic treatments, suggesting that miR-454-3p could bind from the 205th to 211th bases of circGNAI2 (Fig. 4F). We found that the expression of miR-454-3p was upregulated in both TNBC cells and tissues when compared to that in adjacent normal tissues and cell lines (Fig. 4, G and H), and the expression of miR-454-3p was negatively correlated with circGNAI2 expression in TNBC tissues (Fig. 4I).

Figure 4.

CircGNAI2 serves as a sponge for miR-454-3p.A, Venn diagram showing the potential target miRNAs of circGNAI2. B-C, RT-qPCR was conducted to evaluate the expression of miRNAs in MDA-MB-231 and HCC-1937 cell lines transfected with circGNAI2 overexpression vectors, with empty vector-transfected cells serving as negative controls. D, putative binding sites between circGNAI2 and miR-454-3p were predicted. Plasmids containing WT or MUT sequences of the putative binding sites were constructed. E, the enrichment of circGNAI2 was detected after immunoprecipitation with AGO2 in RIP assay. F, dual luciferase reporter assay demonstrated that miR-454-3p is a direct target of circGNAI2. G-H, the expression of miR-454-3p was upregulated in TNBC tissues and cell lines compared with adjacent normal tissues and cell lines. I, a negative correlation between the expression of circGNAI2 and miR-454-3p was found with Pearson correlation analysis in TNBC tissues. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

CircGNAI2 suppresses TNBC progression possibly via the miR-454-3p/VGLL4/STAT3 axis

VGLL4 is a transcriptional cofactor of VGLL family, which has been confirmed as a novel tumor suppressor in TNBC. Our previous study demonstrated that miR-454-3p exerts an oncogenic role in TNBC cells and that VGLL4 is a direct target of miR-454-3p. In addition, VGLL4 inhibits STAT3 phosphorylation and its downstream transcriptional activity by interacting with STAT3 protein (18).

Based on the relationship between circGNAI2, miR-454-3p, and VGLL4, we hypothesized that circGNAI2 might regulate the expression of VGLL4 through miR-454-3p. Western blotting results showed that silencing of circGNAI2 resulted in a downregulation of the VGLL4 protein level (Fig. 5, A–C), while overexpression of circGNAI2 promoted the expression of VGLL4 (Fig. 5, D–F). Interestingly, we found that circGNAI2 expression positively correlated with VGLL4 mRNA levels (measured by RT-qPCR) in TNBC tissues (Fig. 5G). To confirm that circGNAI2 inhibits TNBC progression via the miR-454-3p/VGLL4 axis, we performed rescue experiments by co-transfecting circGNAI2 and miR-454-3p mimics into MDA-MB-231 and HCC-1937 cells. We found that miR-454-3p mimics partially abolished the inhibitory effect of circGNAI2 on TNBC cell proliferation and migration (Fig. 6, A–D). Similarly, miR-454-3p mimics partially reversed the upregulation of VGLL4 protein levels and downregulation of p-STAT3 protein levels induced by circGNAI2 overexpression (Fig. 6, E–G). In addition, to disrupt the interaction between circGNAI2 and miR-454-3p, we constructed circGNAI2-mutant type (circGNAI2-MT), as well as circGNAI2-wild type (circGNAI2-WT) and empty vector control (NC). Cells were transfected with circGNAI2-WT, circGNAI2-MT, and NC, followed by functional assays. Notably, circGNAI2-WT, but not the circGNAI2-MT, significantly inhibited cellular proliferation and migration in MDA-MB-231 and HCC-1937 cells (Fig. S2, A and B). Mechanistically, circGNAI2-WT upregulated VGLL4 while downregulating p-STAT3 levels, whereas circGNAI2-MT showed no significant effects on these pathway proteins (Fig. S2, C and D). In summary, we confirmed that circGNAI2 suppressed the progression of TNBC, which was associated with VGLL4 upregulation and STAT3 phosphorylation inhibition via miR-454-3p.

Figure 5.

CircGNAI2 promotes VGLL4 expression in TNBC cells.A-C, the protein level of VGLL4, STAT3 and p-STAT3 were evaluated in MDA-MB-231 and HCC-1937 cell lines transfected with si-circGNAI2. D-F, the protein level of VGLL4, STAT3 and p-STAT3 were evaluated in MDA-MB-231 and HCC-1937 cell lines transfected with circGNAI2. G, Pearson correlation analysis revealed a positive correlation between circGNAI2 and VGLL4 expression in TNBC tissues. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Figure 6.

CircGNAI2 suppresses TNBC progression via miR-454-3p/VGLL4 axis.A-B, miR-454-3p-mimics rescued the suppressive effects of circGNAI2 on proliferation in MDA-MB-231 and HCC-1937 cell lines by colony formation assay. C-D, miR-454-3p-mimics rescued the suppressive effects of circGNAI2 on migration in MDA-MB-231 and HCC-1937 cell lines by colony transwell assay. E-G, Western blot showed that miR-454-3p mimics can partly rescue the high expression of VGLL4 and the low expression of p-STAT3 caused by circGNAI2 in MDA-MB-231 and HCC-1937 cell lines. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Overexpression of circGNAI2 suppresses TNBC tumor growth in vivo

To further clarify the tumor suppressor role of circGNAI2 in TNBC, we performed a xenograft assay to determine the biological function of circGNAI2 in vivo using MDA-MB-231 cells stably overexpressing circGNAI2 (Fig. 7A). MDA-MB-231 cells stably overexpressing circGNAI2 or control cells were injected into the second mammary fat pad of mice in both the groups. Six weeks later, the tumors dissected from the sacrificed mice were photographed (Fig. 7B). By analyzing the volume and weight of the tumors in the two groups, we found that circGNAI2 inhibited the progression in vivo (Fig. 7, C and D). Subsequently, we demonstrated that the protein expression of VGLL4 was upregulated and that of p-STAT3 was downregulated in the tumors of the circGNAI2-overexpression group by western blotting and IHC (Fig. 7, E and F).

Figure 7.

Overexpression of circGNAI2 suppresses TNBC tumor growth in vivo.A, overexpression of circGNAI2 was confirmed by RT-qPCR in MDA-MB-231. B, representative images of xenograft tumors in nude mice (6 mice per group). C, average tumor volume of nude mice. D, average tumor weight of nude mice. E, extract protein from tumors and measuring the expression of VGLL4, STAT3 and p-STAT3 by Western blot. F, immunohistochemistry (IHC) staining of VGLL4 and p-STAT3 in xenografts. ∗∗∗∗p < 0.0001.

EIF4A3-regulated circGNAI2 expression in TNBC cells

According to the Circular RNA Interactome prediction, we found that EIF4A3 had four binding sites in the upstream and downstream regions of the GNAI2 pre-mRNA transcript (Fig. 8A). Accordingly, we performed RIP assay with an anti-EIF4A3 antibody and observed significant enrichment of GNAI2 pre-mRNA in the EIF4A3 immunoprecipitates compared to the negative control, suggesting that that EIF4A3 could bind to the upstream flanking sequence of GNAI2 pre-mRNA (Fig. 8B). To elucidate the regulatory function of EIF4A3 in circGNAI2 biogenesis, we generated isogenic TNBC cells with EIF4A3 knockdown or overexpression. Knockdown of EIF4A3 markedly elevated circGNAI2 abundance but reduced linear GNAI2 mRNA levels in MDA-MB-231 and HCC-1937 cells (Fig. 8C). In contrast, EIF4A3 overexpression diminished circGNAI2 production while promoting linear GNAI2 expression (Fig. 8D). These findings demonstrate that EIF4A3 negatively regulates circGNAI2 expression in TNBC. Additionally, the correlation analysis demonstrated that EIF4A3 expression was not significantly correlated with circGNAI2 levels in TNBC samples (r = −0.189, p = 0.208, Pearson correlation) (Fig. S1C), this observation may result from competitive molecular interference, spatiotemporal expression heterogeneity or other underlying mechanisms. Taken together, our results suggest that EIF4A3-regulated circGNAI2 inhibits the progression of TNBC, likely through the miR-454-3p/VGLL4/STAT3 axis (Fig. 8E).

Figure 8.

EIF4A3 regulates circGNAI2 expression in TNBC cells.A, the binding sites of EIF4A3 in the flanking sequences of the GNAI2 transcript were predicted. B, RIP assay was performed to validate the binding of EIF4A3 to the upstream flanking sequence of GNAI2 pre-mRNA. IgG were used as the negative controls. C-D, the level of circGNAI2 and GNAI2 mRNA were evaluated in MDA-MB-231 and HCC-1937 cell lines with EIF4A3 knockdown or overexpression. E, schematic diagram illustrating the mechanism of EIF4A3-regulated circGNAI2 sponging miR-454-3p to regulate the VGLL4-STAT3 signaling axis in TNBC. ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Discussion

Although a plethora of new therapies and medicine has been incorporated into breast cancer treatment to tackle this public-health issue on a global scale in women (19, 20), the clinical prognosis for TNBC remains poor due to lacking effective therapeutic targets. Therefore, the treatment of TNBC remains challenging. Therefore, it is imperative to explore novel therapeutic targets for TNBC and to develop drugs based on specific targets.

Over the past decades, emerging evidence has demonstrated that non-coding RNAs (ncRNAs) participate in tumorigenesis and progression of multiple cancers, including BC (21, 22). The application of ncRNAs as biomarkers or intervention targets can provide new insights into the diagnosis and treatment of tumors (23, 24). Different from other linear non-coding RNAs, circular RNAs (circRNAs) are a group of specific highly conservative and stable RNAs with covalently closed cyclic structure (25). Furthermore, the detectability and high tissue specificity of circRNAs in liquid biopsy samples has strong potential as diagnostic, prognostic, and predictive biomarkers (26).

In this study, we found that circGNAI2 had lower expression in TNBC tissues and cell lines and was negatively associated with lymph node metastasis and TNM tumor stage, suggesting that it may be used as an indicator of poor prognosis in TNBC. Subsequently, functional in vivo and in vitro experiments confirmed that circGNAI2 significantly inhibits the proliferation and migration of TNBC cells. It should be noted that this study has certain limitations, the potential effects of circGNAI2 on cell death or metastasis require further investigation, larger-scale studies are needed to validate the clinical implications of differential circGNAI2 expression in TNBC.

Typically, the mechanism of circRNAs is associated with their subcellular localization, which regulates the transcription of their parental genes in the nucleus, whereas most cytoplasmic circRNAs serve as miRNA sponges (14). Since circGNAI2 is mainly located in the cytoplasm of TNBC cells, we explored its potential role in sponging miRNAs and screened and characterized the miRNAs directly downstream of circGNAI2 and miR-454-3p. High miR-454-3p expression is associated with lower disease-free and overall survival in patients with TNBC (27). Our previous study also suggested that miR-454-3p exerts an oncogenic role and that VGLL4 is a direct target of miR-454-3p in TNBC cells (18). Here, we revealed that circGNAI2 is involved in tumor suppression of TNBC by specific sponge-binding of miR-454-3p and upregulation of the target gene VGLL4.

VGLL family proteins are a group of transcriptional cofactors with four mammalian VGLL proteins, named VGLL1-4 (28), which are involved in BC tumorigenesis and progression. For instance, VGLL1 was discovered to be a driver of proliferation and invasion (29), and VGLL3 inhibits ER-positive breast cancer cells by recruiting the repressor to the super-enhancer of the ESR1 gene, leading to the transcriptional silencing of ESR (30). However, VGLL4 is functionally different, because it contains two TDU motifs (31). In lung cancer cells, VGLL4 negatively regulates the YAP-TEAD transcriptional complex and inhibits tumor formation (28). In colorectal carcinoma cells, VGLL4 inhibits tumorigenesis by targeting the TEAD4-TCF4 complex and suppressing TCF4 (32). In our previous studies, VGLL4 functioned as a tumor suppressor in TNBC by interacting with STAT3 and inhibiting STAT3 phosphorylation (18). Together, these findings highlight the central importance of VGLL4 in repressing tumorigenesis and suggest potential strategies for modulating VGLL4 expression in cancer.

Based on the important role of circGNAI2 in upregulating the critical tumor suppressor VGLL4, we further explored the regulatory mechanism of circGNAI2 expression. Through bioinformatics prediction, we found EIF4A3 that is a core component of the exon junction complex (EJC) and is important in RNA splicing, trafficking, translation, and degradation (33), has four binding sites on both the upstream and downstream regions of the GNAI2 pre-mRNA transcript. EIF4A3 is transcriptionally overexpressed in various malignancies, including breast cancer, lung cancer, and urinary tract tumors. Notably, high EIF4A3 expression correlates with poor overall survival in breast cancer patients (34). Functionally, EIF4A3 promotes breast cancer cell proliferation (35), and enhances EMT (epithelial-mesenchymal transition) as well as invasion capacity in glioma by upregulating Snail and Twist expressions (36). Furthermore, EIF4A3 promotes tumor progression by activating the PI3K/AKT signaling pathway in TNBC (10). The steady-state abundance of circRNAs is co-regulated by their biogenesis, nuclear export, and degradation (14). For example, the SF3B complex, a core component of U2 small nuclear ribonucleoproteins (snRNPs), recognizes branch point sequences and 3′ splice sites. Disruption of the splicing factor SF3b1, alters RNA processing at the Laccase2 gene, resulting in elevated circRNA levels coupled with reduced linear RNA expression (37). EIF4A3 exhibits bidirectional regulation of circRNA biogenesis, serving as either a positive regulator or a suppressor. For instance, EIF4A3 promotes circRNA circularization through the exon junction complex (38, 39). Conversely, it has been demonstrated to inhibit circ_0087429 expression by binding to the flanking regions of SPIN1 pre-mRNA (40). Our results suggested that EIF4A3 promoted linear GNAI2 mRNA splicing and suppressed circGNAI2 biogenesis by modulating GNAI2 pre-mRNA processing.

In this study, we found that circGNAI2, regulated by EIF4A3, suppressed TNBC progression possibly via the miR-454-3p/VGLL4/STAT3 axis and could potentially be used as a targeted therapeutic or prognostic biomarker for TNBC.

Experimental procedures

Clinical samples

This study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Institutional Ethics Committee of the Affiliated Hospital of Qingdao University (Project Number: QYFY-WZLL-27362). 46 pairs of matched TNBC and adjacent normal tissues were obtained from patients who had undergone surgery at the Breast Disease Center of the Affiliated Hospital of Qingdao University. Patients who had received preoperative radiotherapy or chemotherapy were excluded from the study. All tissue samples were stored in liquid nitrogen immediately after resection, until further use.

Cell culture and transfection

All BC cell lines (MDA-MB-231, BT-549, HCC-1937, MDA-MB-468, HS578T, MCF7, T47D and SK-BR-3), normal breast epithelial cell line (MCF-10A), and HEK-293T cells were acquired from the Chinese Academy of Sciences. All the cells were cultured in Dulbecco’s Modified Eagle’s medium (DMEM; Gibco, USA) supplemented with 10% Fetal Bovine Serum (FBS) and 1% penicillin-streptomycin (PS) in a 5% CO2 incubator at 37 °C. Small interfering RNA targeting circGNAI2 (si-circGNAI2-1 and si-circGNAI2-2) and the negative control (si-NC) were purchased from IBSbio. The inhibitors (anti-miR-454-3p), mimics (miR-454-3p), and their respective negative controls (anti-miR-NC and miR-NC) for miR-454-3p were purchased from RiboBio (Guangzhou). Lentiviral plasmids overexpressing circGNAI2 (circGNAI2) and a negative control (NC) were designed by ZORIN Biotechnology Co., Ltd. EIF4A3 knockdown was performed using siRNA (Shanghai Jierui Bioengineering), and overexpression was achieved with an EIF4A3-expressing plasmid (Wuhan Miaoling Biotechnology, China). Lipofectamine 2000 (Invitrogen, USA) was used for transfection, according to the manufacturer’s instructions.

Quantitative real-time polymerase chain reaction (qRT-PCR)

The cDNA of tissues and cells were obtained using TRIzol reagent (Invitrogen) and Hifair III first Strand cDNA Synthesis SuperMix (Yeasen, China). Hieff qPCR SYBR Green Master Mix was used for qRT-PCR. 18S rRNA, U6, and GAPDH were used as internal controls for circRNAs, miRNAs, and mRNA, respectively. Data were quantified using the 2^-ΔΔCt method. Primers used in this study are listed in Table S1.

Confirming specificity for circGNAI2

Polymerase chain reaction (PCR) was performed using 2×Hieff Robust PCR Master Mix (YEASEN), and the PCR products amplified using the circGNAI2 primers were separated on a 1% agarose gel. The gel was scanned using the Gel Doc XR + imager (Bio-Rad).

Ribonuclease R (RNase R) treatment and actinomycin D assay

For enzyme inactivation, RNAs extracted from cells were treated with RNase R at 37 °C for 30 min. For actinomycin D assay, MDA-MB-231 and HCC-1937 cells were treated with 2 μg/ml actinomycin D (Merck) to block transcription at 0 h, 4 h, 8 h, and 12 h respectively. The remaining RNAs extracted from the treated cells were assessed by qRT-PCR.

Subcellular fraction

The Thermo Invitrogen PARIS Kit (Invitrogen) was used for subcellular fractionation, according to the manufacturer’s instructions. U6 and GAPDH were used as the nuclear and cytoplasmic controls, respectively.

Fluorescent in situ hybridization (FISH)

A specific probe for circGNAI2 for FISH was designed and synthesized by RiboBio. The Ribo Fluorescent In Situ Hybridization Kit (Ribo) was used to detect the localization of circGNAI2. The nuclei were stained with 4′,6-Diamidino-2-Phenylindole (DAPI). A fluorescence microscope (Leica) was used for the image acquisition.

MTT assay

TNBC cells were seeded into 96-well plates at a density of 2000 cells per well in 200 μl medium. 20 μl MTT reagent (YEASEN, China) were added to each well at 0 h, 24 h, 48 h, 72 h and 96 h after seeding. After incubation for 4 to 6 h, the supernatant was replaced with 150 μl DMSO (Sangon, China). Optical density (OD) at 490 nm was measured using a microplate spectrophotometer (BioTek, Germany).

Colony formation assay

TNBC cells were seeded in 6-well plates at a density of 1000 cells per well. After incubation for approximately 10 days, the colonies were visible. The cell colonies were subsequently washed with phosphate-buffered saline (PBS), fixed with 95% ethanol, and stained with 0.1% crystal violet. Representative photographs were taken and the number of colonies was counted.

Wound healing assay

TNBC cells were seeded in 6-well plates. When the cells reached approximately 95% confluence, a scratch was made on the surface of the cell monolayer with a 200 μl pipette tip. The cells were then cultured in DMEM supplemented with 2%FBS. Wound healing at the same location was observed and assessed under a microscope at 0 h, 24 h, 36 h and 48 h.

Transwell assay

TNBC cells were added to the upper chamber with 200 μl of serum-free medium, and medium containing 10% FBS was added to the lower chamber. After culturing for 12 h to 16 h for MDA-MB-231 and HCC-1937 cell lines, the cells that migrated to the opposite side of the filter were fixed, stained, photographed (Leica Microsystems, Germany) and counted.

Dual-luciferase reporter assay

Wild-type (WT) and mutant-type (MUT) reporter plasmids of circGNAI2 were constructed according to the predicted sequence of the binding sites with miR-454-3p (IBSbio). MiR-454-3p mimics or miR-454-3p NC were cotransfected with WT or MUT reporter plasmids into HEK-293T cells. After incubation for approximately 48 h, the total protein from the cell lysates was collected and centrifuged, and a dual-luciferase reporter assay kit (YEASEN, China) was used to detect luciferase activity. The ratio of firefly luciferase to Renilla luciferase was calculated.

RNA immunoprecipitation assay (RIP)

A BersinBioTM RNA Immunoprecipitation Kit (BersinBio) was used to perform the RIP assay, according to the manufacturer’s instructions. Anti-Ago2 (Abclonal), anti-EIF4A3 (Proteintech), and anti-IgG (Abclonal) antibodies were used for immunoprecipitation. The enriched RNAs were further analyzed using RT-qPCR.

Western blot

Total protein was extracted by RIPA lysis buffer (Beyotime) together with PMSF (Beyotime). After separation on 10% sodium dodecyl sulfate-polyacrylamide (SDS-PAGE) gels, proteins were transferred to nitrocellulose membranes (Beyotime). Membranes were blocked with 5% non-fat milk for 1 h at room temperature and then immunoblotted overnight at 4 °C with the following primary antibodies: anti-VGLL4 (1:1000, Abclonal), anti-STAT3 (1:1000, Proteintech), anti-phospho-STAT3(Tyr705) (1:1000, Proteintech), and anti-β-actin (1:10,000, Abclonal). After incubation with diluted secondary antibodies for 1 h at room temperature, the bands were scanned and analyzed using an Odyssey Infrared scanning system (LI-COR Biosciences).

Xenografts experiment

4-week-old female BALB/c nude mice were obtained from SLAC and divided into two groups randomly (n = 6, each group). 1 × 10⁶ transfected MDA-MB-231 cells with stably expressed circGNAI2 or NC were injected into the second mammary fat pad of mice in both the groups. Six weeks later, all mice were sacrificed by cervical dislocation, and the collected tumor volume was measured and calculated as follows: volume (mm³) = width² × length/2. All experimental procedures were approved by the Animal Ethics Committee of the Affiliated Hospital of Qingdao University (Project Number: AHQU20210702).

Immunohistochemistry (IHC)

Tumor tissues collected from BALB/c nude mice were fixed in 4% paraformaldehyde, dehydrated using an ethanol solution, embedded in paraffin, and sectioned into 4 μm slides. The slides were then incubated with anti-VGLL4 and anti-p-STAT3(Tyr705) (Proteintech, USA). Images were captured using Leica Microsystems.

Statistical analysis

Data obtained from at least three independent experiments were analyzed using GraphPad Prism (v8.3.0). Data are presented as the mean ± standard deviation (SD) and were considered significant at p-values <0.05. Relationships between the expression of circGNAI2 and various clinicopathological variables were analyzed using the chi-squared test and Fisher's exact test. Comparisons between paired specimens were analyzed using the Wilcoxon matched-pair signed-rank test. Unpaired samples were analyzed using unpaired Student’s t test.

Data availability

The data used to support the findings of this study are included in the article and its additional files.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Paul Shapiro