Transcriptional factor six2 promotes the competitive endogenous RNA network between CYP4Z1 and pseudogene CYP4Z2P responsible for maintaining the stemness of breast cancer cells

Paraffin-embedded breast cancer tissue samples were obtained in our previous work [26]. Eight pairs of fresh breast cancer and normal adjacent tissues were collected from Huai An First People’s Hospital and Jiangsu Province Hospital of TCM between September 2017 and January 2018. Written informed consent from all patients and approval of the hospital ethics review committees were obtained. Online clinical posited data (http://​www.​firebrowse.​org/), Breast Cancer Gene-Expression Miner (version 4.0; http://​bcgenex.​centregauducheau​.​fr/), and R2: Genomics Analysis and Visualization Platform (http://​hgserver1.​amc.​nl/​cgi-bin/​r2/​main.​cgi) were used for the analyses of gene expression and correlation. The diagnostic values of the genes were analyzed by KM plotter (http://​kmplot.​com) to obtain KM survival plots in which the number at risk is indicated below the main plot [27]. Hazard ratio, 95% confidence intervals, and log-rank P values were calculated and displayed on the webpage.

The human breast cancer cell lines MCF-7, MDA-MB-231, and HEK293T were preserved in our laboratory. Adriamycin-resistant MCF-7-Adr cells were purchased from KeyGen BioTECH (Nanjing, China). The cell line was authenticated every year through short tandem repeat (STR) DNA profiling. HEK293T and MCF-7 cells were cultured in DMEM (Gibco, Grand Island, NY, USA), MCF-7-Adr cells were cultured in 1640 medium (Gibco), and MDA-MB-231 cells were cultured in L-15 medium (Gibco) at 37 °C under a humidified atmosphere with 5% CO₂. All of the media were supplemented with 10% FBS (Gibco), 80 U/ml penicillin, and 0.08 mg/ml streptomycin. PI3K inhibitor (LY-294002) and ERK1/2 inhibitor (VX-11e) were purchased from APExBIO. Adriamycin was purchased from Zhongda Hospital Southeast University.

Total RNA from the cells was extracted using TransZol Up (Cat. No. ET111-01, TransGen Biotech, Beijing, China) following the manufacturer’s recommendation. Total RNA from paraffin-embedded breast cancer tissues was extracted using a total RNA extraction kit for paraffin-embedded tissues (Cat. No. DP439, TianGen Biotech, Beijing, China) according to standard protocols. Then, complementary DNA (cDNA) was reverse-transcribed using M-MLV (H-) Reverse Transcriptase (Cat. No. R021-01, Vazyme, Nanjing, China) according to the manufacturer’s protocol. qRT-PCR was performed with AceQ Universal SYBR qPCR Master Mix (Cat. No. Q511-02, Vazyme). A melting curve analysis was performed routinely to check the amplification specificity. cDNA templates were analyzed in triplicate, and GAPDH was used as an internal control. The relative expression level of each transcript was calculated by the 2^-△△ct method. The qRT-PCR primers are described in Additional file 1: Table S1.

The detailed procedure was described in our previous study [26]. Protein in fresh tissues was extracted using total protein extraction kit (Invent, USA) following the manufacturer’s recommendation. β-actin or GAPDH was used as an internal reference. Detailed information on the antibodies used in this work is given in Additional file 2: Table S2.

CD24 and CD44 expression was analyzed in cells derived from monolayer cultures following dissociation in trypsin-EDTA at 37 °C. At least 1 × 10⁶ cells were pelleted by centrifugation at 300×g and 4 °C for 5 min. Then, cells were washed in PBS, re-suspended with anti-CD24-PE (BD Biosciences, USA) and anti-CD44-APC (BD Biosciences, USA), and then incubated at 4 °C for 30 min in the dark. The labeled cells were washed using PBS and analyzed using a flow cytometer (BD, USA). The negative fraction was determined using appropriate isotype controls.

A chromatin immunoprecipitation (ChIP) assay was performed using the EZ-Magna ChIP™ A/G Chromatin Immunoprecipitation Kit (Cat. No. 17–10086, Merck) following the manufacturer’s protocols. Primers flanking the six2 binding sites on the promoters of CYP4Z1 and pseudogene CYP4Z2P were used for qRT-PCR. The sequences of the primers for ChIP analysis were denoted in Additional file 3: Table S3.

ChIP-sequencing analysis was performed by GENEWIZ (Suzhou, China). ChIP-seq raw reads were aligned to a human reference genome (hg19) using cutadapt (version 1.9.1) to pass filter data and acquire clean data. Up to mismatch per read was allowed. The quality of the sequencing data was assessed using FastQC (v0.10.1), and only uniquely mapped reads were kept for downstream analysis. The data are available in the Gene Expression Omnibus (GEO) database as GSE117145.

RNA sequencing and data analysis were conducted by Novogene (Beijing, China). The data are available in the Gene Expression Omnibus (GEO) database as GSE116984.

A tissue microarray including 30 breast cancer tissues and 30 normal adjacent tissues was purchased from OUTDO IVD (Shanghai, China). Further immunohistochemistry was performed to detect the expression of six2 following the protocols described in our recent work [28].

When cells confluency reached 50%, cells were transfected with a final concentration of 50 nM small interfering RNA (siRNA) or 50 nM NC (si-NC, inhibitor-NC, mi-NC) which were synthesized by Biomics Biotechnology (Nantong, China) using jetPRIME (Polyplus Transfection, France) following the manufacturer’s protocols. The siRNA sequences are mentioned in Additional file 4: Table S4.

A mammosphere formation assay was performed using MammoCult™ Human Medium Kit (STEMCELL Technologies, Canada). A total of 3 × 10³ cells were mixed with 500 μl of Complete MammoCult™ Medium in the presence of 4 μg/ml Heparin (STEMCELL Technologies, Canada) and 0.48 μg/ml hydrocortisone (STEMCELL Technologies, Canada) and seeded in 24-well ultra-low attachment plates (Corning, USA) for 7 days. The spheres were counted and photographed. All images were obtained with a Leica DMI microscope (DE).

To construct stable expression cells, the six2 coding area or the CYP4Z1 3′UTR or CYP4Z2P 3′UTR sequences were subcloned into pLVX-ZsGreen. siRNA oligos were purchased from GenePharma (Shanghai, China). After annealing, double-strand oligos were inserted into the lentiviral pLKO.1-puro vector (Addgene). To package lentivirus, HEK293T cells were co-transfected using Lentifectin (ABM, USA) with the lentiviral vector and packaging vectors psPAX2 and pMD2.G. MCF-7 and MDA-MB-231 cells were infected with the virus in the presence of 2 μg/ml polybrene. Cells infected with Plko.1-derived vectors were selected with puromycin (Sigma, 2 μg/ml) for 2 weeks. Cells infected with pLVX-ZsGreen-derived vectors were selected by fluorescent cell sorting. Western blot and qRT-PCR analyses were used to verify expression levels. MCF-7 cells stably overexpressing CYP4Z1 3′UTR, CYP4Z2P 3′UTR, and six2 were designated as MCF-7-Z1-UTR, MCF-7-Z2P-UTR, and MCF-7-six2, respectively. MCF-7 cells stably knocking down CYP4Z1 3′UTR, CYP4Z2P 3′UTR, and six2 stable were denoted as MCF-7-Plko-Z1, MCF-7-Plko-Z2P, and MCF-7-Plko-six2, respectively. MDA-MB-231 cells stably overexpressing six2 were designated as 231-six2, while those stably knocking down CYP4Z1 3′UTR, CYP4Z2P 3′UTR, and six2 were denoted as 231-Plko-Z1, 231-Plko-Z2P, and 231-Plko-six2, respectively. The sequences of the primers used for plasmid constructions were listed in Additional file 5: Table S5.

The promoter sequences of CYP4Z1 (1–2403 bp) and the pseudogene CYP4Z2P (1–2998 bp) were inserted into the pGL3-promoter vector, and truncations of these sequences were also inserted into the pGL3-promoter vector. Then, potential six2 binding sites were mutated using the Fast Mutagenesis Kit V2 (Vazyme, China) following the manufacturer’s instructions and inserted into the pGL3-promoter vector as well. All of the abovementioned constructs were verified by DNA sequencing before use. To analyze the activity of the abovementioned constructs, they were individually co-transfected with β-gal and a six2 overexpression vector into MCF-7 cells. Seventy-two hours later, the luciferase activity was measured by a POLARstar Omega multimode microplate reader according to the manufacturer’s protocol and normalized to β-gal. All primer sequences for this experiment are listed in Additional file 6: Table S6.

Cell viability was assessed using MTT (Amresco, USA) staining. Cells were seeded at 3 × 10³ cells/well in 96-well cell plates overnight and then treated with different concentrations of adriamycin for 48 h. During the last 3.5 h, the cells were exposed to MTT (5 mg/ml) and the resulting formazan crystals were dissolved in 150 μl of DMSO and measured using a spectrophotometer(BIO-RAD) at a test wavelength of 490 nM. The experiments were conducted in triplicate.

The procedure was referenced in our previous work [28]. Briefly, six-week-old male athymic BALB/c nude mice were purchased from the Model Animal Research Center of Nanjing University, housed and fed under standard pathogen-free conditions. For the tumor-limiting dilution assay, 1 × 10⁶ and 1 × 10⁵ MCF-7 cells, 1 × 10⁵ and 1 × 10⁴ MDA-MB-231 cells, or 1 × 10⁵ MCF-Adr cells receiving different treatment were orthotopically implanted in the inguinal mammary gland of mice. On day 8, all the mice were euthanized and tumor tissues were collected and weighed. The stem cell frequencies were calculated using an ELDA (http://​bioinf.​wehi.​edu.​au/​software/​elda/) [29]. All the animal experiments were carried out with the approval of the Ethics Committee for Animal Experimentation of China Pharmaceutical University.

GraphPad Prism 5.01 software (GraphPad Software, Inc., La Jolla, CA, USA) was used for the statistical analysis. All the data were obtained from at least three independent experiments and are presented as the means ± standard deviations (SDs). Datasets with only two groups were analyzed using Student’s t test. Differences between multiple groups were analyzed by one-way analysis of variance with the Tukey-Kramer post hoc test. P < 0.05 was considered indicative of a statistically significant difference.

The expression levels of CYP4Z1 and its pseudogene CYP4Z2P were initially examined in breast tumor and normal adjacent tissues via online clinical deposited data (http://​www.​firebrowse.​org/). As shown in Fig. 1a, b, the expression of CYP4Z1 and the pseudogene CYP4Z2P was significantly increased in breast tumor tissues, which is consistent with our previous work in which CYP4Z1 and pseudogene CYP4Z2P expression was detected in clinical samples [26, 30]. KM-plotter analysis (http://​kmplot.​com) indicated that CYP4Z1 expression was negatively correlated with the OS of breast cancer patients (Fig. 1c). We defined the CYP4Z1-3′UTR- or CYP4Z2P-3′UTR-regulated transcriptome in MCF-7 cells with CYP4Z1- or CYP4Z2P-3′UTR overexpression and subsequently performed expression profiling, and the results revealed a substantial and highly statistically significant overlap between genes regulated by both CYP4Z1-3′UTR and CYP4Z2P-3′UTR in MCF-7 cells (Fig. 1d). Additionally, CYP4Z1-3′UTR and CYP4Z2P-3′UTR activated (3854 vs. 3788) or repressed (3404 vs. 3934) similar numbers of genes in MCF-7 cells (Fig. 1e, f). Further gene set enrichment analysis (GSEA) of this dataset revealed that a negative enrichment of stem cell-differentiated signatures was observed in MCF-7 cells with CYP4Z2P-3′UTR overexpression (Fig. 1g), and the top positive signatures associated with CYP4Z1-3'UTR overexpression were the embryonic stem cell function and adult tissue stem modules (Fig. 1h, i) [31], both of which are indicative of a stemness phenotype. Functional annotation analysis revealed that the overexpression of CYP4Z2P-3′UTR or CYP4Z1-3′UTR activated signaling pathways regulating the pluripotency of stem cells (Fig. 1j). Among the differentially expressed genes, we identified a set of gene signatures related to epithelial cancer stem cells in MCF-7 cells with CYP4Z2P-3′UTR and CYP4Z1-3′UTR overexpression, including YY1, KRAS, YAP1, and HMGB2 (Fig. 1k) [31]. Consistent with the activation of stem cell function, we detected an increase in a set of cell cycle-related genes, namely CDK5R1, CDK1, CDK2, and CDK7 (Fig. 1). Notably, the expression of CYP4Z1 and CYP4Z2P displayed a positive correlation in clinical breast cancer samples (P < 0.001, Fig. 1l).

As previous studies have demonstrated that non-adherent spheres are highly enriched for CSCs [32, 33], the expression levels of CYP4Z1 and its pseudogene CYP4Z2P were examined in non-adherent MCF-7 spheres and parental cells. qRT-PCR results showed that non-adherent MCF-7 spheres displayed significantly higher levels of CYP4Z1 and CYP4Z2P relative to monolayer cultures of MCF-7 cells (Additional file 7: Figure S1A). We then determined whether the overexpression of CYP4Z1- or CYP4Z2P-3′UTR conferred stemness upon breast cancer cells in vitro. First, MCF-7 cells with CYP4Z1- or CYP4Z2P-3′UTR stable overexpression or knockdown were subjected to a spheroid formation assay. The infection efficiency was confirmed by qRT-PCR (Additional file 7: Figure S1B and C). Both the sphere size and number were increased in MCF-7 cells with CYP4Z1- or CYP4Z2P-3′UTR overexpression, whereas the knockdown of CYP4Z1- or CYP4Z2P-3′UTR exerted the opposite effects (Additional file 7: Figure S1D). Additionally, the overexpression of CYP4Z1- or CYP4Z2P-3′UTR increased the CD44⁺/CD24⁻ population, which has been identified as having breast CSC markers [34] (Additional file 7: Figure S1E). Moreover, the expression of several pluripotent transcription factors, namely Oct3/4, ALDH1, Nanog, and Sox2, was increased in cells with CYP4Z1- or CYP4Z2P-3′UTR overexpression, whereas the knockdown of CYP4Z1- or CYP4Z2P-3′UTR yielded the opposite effects (Additional file 7: Figure S1F and S1G). These results were recapitulated in MDA-MB-231 cells with or without the knockdown of CYP4Z1- or CYP4Z2P-3′UTR. qRT-PCR analysis confirmed the knockdown efficiency of shRNAs against CYP4Z1- or CYP4Z2P-3′UTR (Additional file 8: Figure S2A and B). As expected, the knockdown of CYP4Z1- or CYP4Z2P-3′UTR resulted in fewer primary mammospheres than control cells (Additional file 8: Figure S2C) and decreased the CD44⁺/CD24⁻ population (Additional file 8: Figure S2D). Additionally, the expression of stemness markers was inhibited in MDA-MB-231 cells with CYP4Z1- or CYP4Z2P-3′UTR knockdown (Additional file 8: Figure S2E and F). Notably, we assessed the expression correlation between Nanog and CYP4Z1 or CYP4Z2P across basal-like breast cancer subtypes using Breast Cancer Gene-Expression Miner (version 4.0; http://​bcgenex.​centregauducheau​.​fr/) and found that Nanog expression is positively correlated with CYP4Z1 or CYP4Z2P expression in basal-like breast cancer subtypes (Additional file 8: Figure S2G and S2H). These results indicate that ceRNET_CC is engaged in tumor stemness in breast cancer.

We further investigated whether ceRNET_CC facilitates the tumor-initiating potential of breast cancer cells in vivo. We compared the capacity of MCF-7 cells with CYP4Z1- or CYP4Z2P-3′UTR overexpression or knockdown to seed tumors at limiting dilutions. Although all of the cell lines could form tumors at a density of 1 × 10⁶ and 1 × 10⁵ cells, CYP4Z1- or CYP4Z2P-3′UTR overexpressed cells showed increased tumor size and weight (Fig. 2a–d) while knockdown of CYP4Z1- or CYP4Z2P-3′UTR decreased the tumor-initiating ability of MCF-7 cells (Fig. 2a–f). Additionally, we performed an in vivo tumorigenic assay with MDA-MB-231 cells after CYP4Z1- or CYP4Z2P-3′UTR knockdown and showed that the knockdown of CYP4Z1- or CYP4Z2P-3′UTR remarkably reduced the tumor-initiating potential of MDA-MB-231 cells (Fig. 2g–k). Therefore, our results demonstrate that ceRNET_CC could promote the stemness of breast cancer cells.

We next sought to explore the mechanisms through which ceRNET_CC promotes the stemness of breast cancer cells. To address this issue, we characterized pathways regulated by CYP4Z1- or CYP4Z2P-3′UTR based on RNA-sequencing data. As expected, we found that the phosphatidylinositol signaling system and MAPK signaling pathways in cancer were some of the most highly upregulated pathways after CYP4Z1- or CYP4Z2P-3′UTR overexpression (Fig. 3a, b). These results were supported by our previous studies showing that ceRNET_CC could act as a subceRNA network for hTERT and promote hTERT expression, thus activating hTERT/PI3K/Akt and ERK1/2 signaling [13, 26]. We then aimed to detect whether ceRNET_CC indeed facilitated the stemness of breast cancer cells through these two pathways. MCF-7-Z1-UTR or MCF-7-Z2P-UTR cells were transfected with siRNA against hTERT (sihTERT) or treated with a PI3K/Akt inhibitor (LY-294002) or ERK1/2 inhibitor (VX-11e), and we then detected the formation of cell spheroids and the CD44⁺/CD24⁻ population. As shown in Fig. 3c, d, the increased cell spheroid formation or CD44⁺/CD24⁻ population induced by CYP4Z1- or CYP4Z2P-3′UTR overexpression was attenuated by sihTERT transfection or by LY-294002 or VX-11e treatment. Additionally, the increased expression of stemness markers induced by CYP4Z1- or CYP4Z2P-3′UTR overexpression was attenuated by sihTERT (Fig. 3e) or by LY-294002 or VX-11e treatment (Fig. 3f) and the knockdown of hTERT or treatment with LY-294002 or VX-11e decreased the p-Akt and p-ERK1/2 levels. Thus, these results demonstrate that ceRNET_CC promotes the stemness of breast cancer cells in a manner dependent on the hTERT/PI3K/Akt and ERK1/2 pathways.

To determine the mechanisms through which ceRNET_CC is regulated, the Genomatix Software Suite v3.10 (https://​www.​genomatix.​de) was used to predict transcriptional factors that could bind to the promoters of CYP4Z1 and CYP4Z2P. We aimed to identify specific transcriptional factors that correlate with tumor progression and tissue development, and six2 attracted our attention because it has been shown to promote breast cancer metastasis [18] and regulate the expansion of the nephron progenitor pool during nephrogenesis, which involves a process similar to CSC formation [35]. We identified one (− 1458 nt) and two putative binding sites (− 1125 nt and − 2716 nt) on the promoters of CYP4Z1 and CYP4Z2P, respectively (Fig. 4a). To confirm the interaction between six2 and CYP4Z1 or CYP4Z2P, MCF-7 cells showing stable overexpression of six2 were constructed by lentivirus infection. We then performed a genome-wide chromatin immunoprecipitation sequencing (ChIP-seq) analysis to identify the six2-bound chromatin regions using an antibody against six2. As shown in Fig. 4b, binding molecular function and developmental process were enriched in six2-overexpressing cells. Importantly, signaling pathways regulating stem cell pluripotency were enriched in six2-overexpressed cells (Fig. 4c). Peak and motif analysis revealed that the 5′-TCAG-3′ motif was highly enriched (P = 1e⁻¹¹) (Fig. 4d), and the majority of six2 peaks were located at the promoters/transcription start sites (Fig. 4e). Importantly, six2 occupied the promoters of CYP4Z1 and CYP4Z2P (Fig. 4f). We subsequently performed a ChIP assay for six2 in MCF-7 cells with or without six2 overexpression. Our results indicated that six2 was indeed bound to the promoters of CYP4Z1 and CYP4Z2P and that six2 overexpression increased six2 binding to these promoters (Fig. 4g, h). Additionally, we constructed luciferase reporter vectors containing different regions of the promoters of CYP4Z1 and CYP4Z2P and found that the luciferase activity of the luciferase reporter vectors containing the putative binding sites was enhanced in MCF-7 cells with six2 overexpression (Additional file 9: Figure S3A–C), whereas the activity of luciferase vectors with mutated binding sites or without putative binding sites was unaffected when the binding sites were mutated (Additional file 9: Figure S3B–D). These results indicated that six2 could directly bind to the 5′-TCAG-3′ motif in the promoters of CYP4Z1 and CYP4Z2P. Consistently, the expression levels of CYP4Z2P and CYP4Z1 were significantly increased or decreased by six2 overexpression or knockdown, respectively (Additional file 9: Figure S3E–H). Notably, a qRT-PCR assay of clinical samples showed that both six2 and CYP4Z1 expression and six2 and CYP4Z2P expression were positively correlated in breast cancer tissues (P < 0.001, Fig. 4i, j). These results suggest that six2 directly binds to the promoters of CYP4Z1 and CYP4Z2P, thus increasing their expression and activating ceRNET_CC.

Notably, we indeed observed a large overlap and positive correlation between gene expression profiles regulated by CYP4Z1-3′UTR- or CYP4Z2P-3′UTR and six2 in MCF-7 cells (Fig. 5a, b). We next explored whether six2 exerts similar effects on ceRNET_CC. First, six2 expression was examined in breast tumor and normal adjacent tissues via online clinical deposited data (http://​www.​firebrowse.​org/). The results showed that six2 expression was significantly increased in breast tumor tissues (Fig. 5c), and a further tissue microarray analysis exhibited consistent results (Fig. 5d). Additionally, an analysis of different clinical samples (http://​hgserver1.​amc.​nl/​cgi-bin/​r2/​main.​cgi) showed that six2 expression was negatively correlated with OS, disease-free survival, and RFS of breast cancer patients (Additional file 10: Figure S4A–F). Notably, we identified a set of gene signatures regulating the pluripotency of stem cells in MCF-7 cells with overexpressing CYP4Z2P-3′UTR, CYP4Z1-3′UTR, and six2, and these genes included AMOTL2, CCNA2, ID1/2, GSK3β, and FGF2 (Fig. 5e). A GSEA revealed that six2 overexpression resulted in the enrichment of gene sets related to embryonic stem cell and mammary stem cell function (Fig. 5f, g), consistent with the CYP4Z1-3′UTR or CYP4Z2P-3′UTR overexpression-mediated changes. As expected, six2 overexpression potentiated sphere formation in MCF-7 cells, whereas the knockdown of six2 exerted the opposite effects (Additional file 11: Figure S5A). Moreover, six2 overexpression increased the CD44⁺/CD24⁻ population, whereas the knockdown of six2 decreased this population (Additional file 11: Figure S5B and S5C). Furthermore, the expression of several pluripotent transcription factors was increased in cells with six2 overexpression and decreased in cells with six2 knockdown (Additional file 11: Figure S5D–G). We further investigated whether six2 could promote the tumor-initiating potential of breast cancer cells in vivo. We compared the capacity of six2-overexpressing or six2-knockdown MCF-7 cells to seed tumors at limiting dilutions. Although all cell lines could form tumors at the densities of 1 × 10⁶ and 1 × 10⁵ cells, the six2-overexpressing cells showed an increase of tumor size and weight, whereas six2-knockdown cells exhibited decreased tumor size and weight (Fig. 6a–d). Furthermore, an in vivo tumorigenic assay was performed with MDA-MB-231 cells after six2 overexpression or knockdown and demonstrated that six2 remarkably attenuated the tumor-initiating potential of MDA-MB-231 cells (Fig. 6e–i). Together, our gain- and loss-of-function assays demonstrate that six2 could promote the stemness of breast cancer cells.

We continued investigating whether the ability of six2 to promote the stemness of breast cancer cells is dependent on ceRNET_CC. MCF-7-six2 cells were transfected with siRNA against CYP4Z1- (si-Z1) or CYP4Z2P-3′UTR (si-Z2P), and the resulting cells were subjected to cell spheroid formation and CD44⁺/CD24⁻ population assays. As shown in Fig. 7a–d, the increased cell spheroid formation, CD44⁺/CD24⁻ population, and stemness marker expression induced by six2 overexpression were nullified by si-Z1 and/or si-Z2P transfection. CYP4Z1- or CYP4Z2P-3′UTR knockdown decreased the hTERT, p-Akt, and p-ERK1/2 levels or even reversed the six2-mediated increase in the hTERT, p-Akt, and p-ERK1/2 levels (Fig. 7e). Notably, co-transfection with si-Z1 and si-Z2P exerted additive effects. Importantly, we constructed MCF-7-six2 cells with CYP4Z1 or CYP4Z2P knockdown, designated as MCF-7-six2-si-Z1 and MCF-7-six2-si-Z2P cells, respectively, and MCF-7-Plko-six2 cells with CYP4Z1- or CYP4Z2P-3′UTR overexpression, denoted as MCF-7-Plko-six2-Z1-UTR and MCF-7-Plko-six2-Z2P-UTR cells, respectively. An in vivo tumorigenic assay showed that six2 overexpression increased the tumor-initiation capacity of MCF-7 cells, an effect that was fully inhibited by CYP4Z1 or CYP4Z2P knockdown, and the decreased tumor-initiation capacity of MCF-7 cells with six2 knockdown was rescued by CYP4Z1- or CYP4Z2P-3′UTR overexpression (Fig. 7f).

Additionally, we detected whether the six2-mediated promotion of breast cancer cell stemness occurs through the downstream effectors of the ceRNA network between CYP4Z1 and the pseudogene CYP4Z2P and the hTERT/PI3K/Akt and ERK1/2 pathways. To address this issue, we characterized pathways regulated by six2 based on RNA-sequencing data. As expected, we found that PI3K-Akt signaling pathway, phosphatidylinositol signaling system, and signaling pathways regulating pluripotency of cells and breast cancer were among the pathways most highly upregulated by six2 overexpression (Additional file 12: Figure S6A). As expected, the increased cell spheroid formation or CD44⁺/CD24⁻ population induced by six2 overexpression was attenuated by sihTERT transfection or by LY-294002 or VX-11e treatment (Additional file 12: Figure S6B–D). Moreover, the increased expression of stemness markers induced by six2 overexpression was attenuated by sihTERT (Additional file 12: Figure S6E), or by LY-294002 or VX-11e treatment (Additional file 12: Figure S6F). Furthermore, the overexpression or knockdown of six2 increased or decreased the expression of hTERT, p-ERK1/2, and p-AKT in MDA-MB-231 cells, respectively (Additional file 12: Figure S6G). Consistent with these results, the knockdown of CYP4Z1, CYP4Z2P, or hTERT or the treatment with LY-294002 or VX-11e reduced the ability of six2 overexpression to promote hTERT expression, PI3K/Akt signaling, ERK1/2 signaling, or the expression of cell stemness markers in MDA-MB-231 cells (Additional file 12: Figure S6H). Notably, we found that the overexpression of CYP4Z1- or CYP4Z2P-3′UTR rescued the six2 knockdown-mediated inhibition of breast cancer cell stemness, characterized by increasing cell spheroid formation (Additional file 13: Figure S7A and S7B), CD44⁺/CD24⁻ population (Additional file 13: Figure S7C), and expression of stemness markers and reactivation of the PI3K/Akt and ERK1/2 signaling (Additional file 13: Figure S7D–G). Thus, these results demonstrate that six2 promotes breast cancer cell stemness dependent on ceRNET_CC.

We have established the ability of the six2/ceRNET_CC axis to promote CSC traits in breast cancer cells. As conferring CSC traits have been confirmed to endow chemoresistance to tumor cells [36], we speculated whether this regulatory axis could decrease adriamycin sensitivity in breast cancer cells by promoting cell stemness. First, the expression of six2, CYP4Z1, and the pseudogene CYP4Z2P was examined in MCF-7 and adriamycin-resistant MCF-7 (MCF-7-Adr) cells via qRT-PCR assay, and the expression was shown to be significantly increased in MCF-7-Adr cells (Fig. 8a). Furthermore, IC₅₀ values were determined in MCF-7 and MDA-MB-231 cells with CYP4Z1- or CYP4Z2P-3′UTR overexpression or knockdown, or with six2 overexpression or knockdown, and we found that overexpression of CYP4Z1- or CYP4Z2P-3′UTR or six2 increased the IC₅₀ values of adriamycin, while knockdown decreased the IC₅₀ values in both MCF-7 and MDA-MB-231 cells (Table 1). MCF-7-Adr cells displayed higher expression of P-gp (a multidrug resistance protein) and pluripotent transcription factors and hyperactivation of PI3K/Akt and ERK1/2 signaling, both of which were attenuated by the knockdown of CYP4Z1- or CYP4Z2P-3′UTR or six2, and the overexpression of CYP4Z1- or CYP4Z2P-3′UTR rescued the attenuation mediated by six2 knockdown (Fig. 8b). Additionally, the knockdown of CYP4Z1- or CYP4Z2P-3′UTR or six2 decreased the spheroid formation ability (Fig. 8c) and the CD44⁺/CD24⁻ population in MCF-7-Adr cells (Fig. 8d), and the decrease induced by six2 knockdown was rescued by the overexpression of CYP4Z1- or CYP4Z2P-3′UTR. Notably, the knockdown of CYP4Z1- or CYP4Z2P-3′UTR or six2 enhanced adriamycin sensitivity, evidenced by the decrease of IC₅₀ values of adriamycin (Table 1). Consistently, the knockdown of CYP4Z1- or CYP4Z2P-3′UTR or six2 impaired the tumor-initiating potential of MCF-7-Adr cells (Fig. 8e, f). Thus, our results indicate that the six2/ceRNET_CC regulatory axis attenuates adriamycin sensitivity by promoting the stemness of breast cancer cells (Additional file 14: Figure S8).