Data retrieval

Bulk RNA-seq data and clinical information for the BC (TCGA-BC) cohort were downloaded from the UCSC Xena browser (https://xenabrowser.net/). Single-cell RNA-seq data for BC (GSE118389) were downloaded from the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/); chip data samples for GSE118389 were obtained from tumor tissues of 6 BC patients, and the sequencing libraries were constructed using an optimized Smart-seq2 method, followed by sequencing on Illumina NextSeq 500. The raw single-cell expression matrix for 1534 cells provided by the GEO database was downloaded. All analyses in this study were conducted using R software, version v3.6.1.

Single-cell RNA-seq analysis

An integration of single-cell transcriptome Figure spectra for downstream analysis was conducted. The R package Seurat (https://satijalab.org/seurat, version 2.2) [11] was employed for the analysis of single-cell RNA-seq and the mitigation of batch effects within the data. The t-SNE algorithm was utilized for non-linear dimensionality reduction of the single-cell sequencing data. Clustering of individual cells, identification of marker genes for each cluster, and exportation of matrices with unique molecular identifier (UMI) values for each gene in individual cells were accomplished using Cell Ranger. Marker genes for various cell clusters were identified through the Seurat package. The majority of the Seurat analyses were conducted using default parameters. For the FeaturePlot function, a max cutoff of 0.5 was employed. Annotation of cell clusters was achieved using the SingleR package [12]. The R package “inferCNV” (https://github.com/broadinstitute/inferCNV) was utilized for the analysis of single-cell copy number variations).

GO enrichment analysis

The ClusterProfiler R package [13] was employed to conduct GO (Gene Ontology) enrichment analysis on differentially expressed genes, with an adjustment made for gene length bias. GO terms were deemed significantly enriched by the differentially expressed genes if they exhibited an adjusted p-value below 0.05.

Survival analysis

The “Survival” R package (https://CRAN.R-project.org/package=Survival) was utilized for survival analysis of risk scores. Wald statistics, generated through univariate Cox proportional hazards regression, could be allocated to each gene as weights. Risk scores for each patient were computed using a linear combination of weighted gene expression (Hazard Ratio, HR) [14]. Overall survival was analyzed through the Kaplan-Meier method. Diagnostic ROC based on the breast cancer dataset from the TCGA database was performed using the “pROC” R package (https://cran.r-project.org/web/packages/pROC/index.html), and visualization of figures was conducted using the “ggplot2” package. The statistical significance of survival differences was tested through the log-rank test. A P-value less than 0.05 was considered to indicate statistical significance.

Clinical sample collection

Biopsy specimens from 35 breast cancer (BC) patients were collected from our institution, with adjacent non-cancerous tissue serving as the normal control. No patients had undergone radiation or chemotherapy prior to surgery. All samples were classified and graded according to the WHO histological classification of breast cancer and confirmed by pathology. The study was approved by our institution’s ethical review committee. Informed consent was obtained from all patients, adhering to the Declaration of Helsinki.

Cell culture and transfection

Normal human mammary epithelial cells MCF-10 A and breast cancer (BC) cell lines T47D and MCF-7 were purchased from the ATCC collection. All the aforementioned cells were cultured in DMEM (10,569,044, Gibco, USA) supplemented with 10% FBS (100,099,141, Gibco, USA) and 1% penicillin-streptomycin (15,070,063, Gibco, USA) and were maintained in an incubator at 37 °C with 5% CO2.

MCF-7 cells, when in the logarithmic phase, were digested with trypsin and then seeded in a 6-well plate at a density of 1 × 105 cells per well. Conventional cultivation was continued for 24 h, and once the cell confluence reached approximately 40%, the cells underwent lentivirus infection according to the lentivirus infection instructions. The cells were divided into three groups: oe-NC group (infected with empty vector control lentivirus, oe-NC), oe-CYP24A1 group (infected with lentivirus oe-CYP24A1), and oe-TFPI2 group (infected with lentivirus oe-TFPI2). Cells were harvested for subsequent experiments 72 h post-transfection.

RT-qPCR

Total RNA from tissues or cells was isolated utilizing Trizol (16,096,020, Thermo Fisher Scientific, New York, USA). For mRNA detection, reverse transcription was executed with a reverse transcription kit (RR047A, Takara, Japan) to produce cDNA. Utilizing the SYBR Premix Ex TaqTM II kit (DRR081, Takara, Japan), sample addition was performed, and qRT-PCR reactions were conducted using a real-time fluorescence quantitative PCR instrument (ABI 7500, ABI, Foster City, CA, USA). GAPDH served as the internal reference gene for coding genes. The PCR program was established as follows: 95 °C for 10 min, followed by 35 cycles of 95 °C for 15 s, 60 °C for 30 s, and 72 °C for 45 s. All qRT-PCR setups were conducted in triplicate. The 2-ΔΔCt method was employed to illustrate the fold-change relationship between the expression of the target gene in the experimental and control groups, using the formula: ΔΔCT = ΔCt experimental group - ΔCt control group, where ΔCt = Ct target gene - Ct reference gene. Ct represents the cycle threshold, or the number of amplification cycles needed for the real-time fluorescence intensity of the reaction to reach a predetermined threshold during the logarithmic phase of amplification. The experiment was repeated thrice. Primer designs are provided in Table S1.

Western blot

Cellular total proteins were extracted using RIPA lysis buffer (P0013B, Beyotime, Shanghai), and the subsequent supernatant was harvested. The BCA kit (P0028, Beyotime, Shanghai) was employed to quantify the total protein concentration of each sample. Following protein denaturation, samples were stored at -80 °C for subsequent use. Depending on the size of the target protein bands, 8-12% SDS gels were prepared, and equal amounts of protein samples were loaded into each lane for electrophoretic separation. The proteins were then transferred onto a PVDF membrane (1,620,177, BIO-RAD, USA) and blocked with 5% BSA at room temperature for 1 h. The primary antibodies, rabbit anti-CYP24A1 (PA5-21704, 1:1000, Thermo Fisher), TFPI2 (ab186747, 1:1000, abcam, UK), and GADPH (ab181602, 1:10000, abcam, UK) were applied and incubated overnight at 4 °C. Following three washes with 1×TBST for 5 min each at room temperature, HRP-labeled goat anti-rabbit IgG (ab6721, 1:5000, Abcam, UK) was added as a secondary antibody and incubated for 1 h at room temperature. After an additional three washes with 1×TBST for 5 min each, the membrane was immersed in ECL reagent (1,705,062, Bio-Rad, USA) for 1 min at room temperature. Excess liquid was removed, and the membrane was enveloped with plastic wrap, followed by band exposure imaging on the Image Quant LAS 4000 C gel imager (GE, USA). GAPDH served as the internal reference for cellular total protein and cytoplasmic protein. The relative protein expression level was quantified by the ratio of the grey value of the target band to the reference band. The expression levels of various proteins were evaluated, with each set of experiments being performed in triplicate.

Immunohistochemistry (IHC)

Breast Cancer (BC) tissues, along with adjacent normal tissues, were fixed using 4% paraformaldehyde, followed by paraffin embedding and serial sectioning to a thickness of 4 μm. Subsequently, the sections were baked at 60 °C for 20 min and transitioned through two xylene baths for 15 min each. The tissues were rehydrated using two 5-minute anhydrous alcohol soaks, followed by immersion in 95%, 90%, 80%, and 70% alcohol solutions, each for 10 min. To block endogenous peroxidase, each section was immersed in 3% H2O2 for 10 min at room temperature. Citrate buffer was added, and sections were microwaved for 3 min for antigen retrieval, followed by a 10-minute room temperature pause. After washing thrice with PBS, the sections were blocked using a normal goat serum solution (Shanghai Bioengineering Co., Ltd., China) for 20 min at room temperature. Primary antibodies, rabbit anti-CYP24A1 (PA5-21704, 1:100, Thermo Fisher) and TFPI2 (ab186747, 1:200, abcam, UK), were applied and sections were incubated at 4 °C overnight. After three washes with PBS, sections were incubated with goat anti-rabbit IgG secondary antibody (ab6721, 1:5000, Abcam, UK) for 30 min. SABC (Vector, USA) was added, and the sections were incubated at 37 °C for 30 min. Subsequent to the addition of a DAB coloring kit (P0203, Beyotime Biotechnology, Shanghai), and a 6-minute coloration period, sections were stained with hematoxylin for 30 s, and gently washed with a slow stream of water. The sections were dehydrated through a sequence of 70%, 80%, 90%, 95% ethanol, and anhydrous ethanol each for 2 min and then cleared in xylene for two 5-minute periods. After sealing with neutral resin, sections were visualized under an upright microscope (BX63, Olympus, Japan). The experiment was replicated three times. Positive staining was determined by the presence of brown or yellow in the cytoplasm, observed and counted across five representative high-power fields.

CCK-8 assay

The cell viability after lentiviral treatment was assessed using a CCK-8 assay kit (K1018, Apexbio, USA). 1 × 104 cells per well were seeded in a 96-well plate (100 µL/well). Subsequently, 10 µL of CCK-8 solution was added at each time point (0 h, 24 h, 48 h, and 72 h), and incubation was performed at 37 °C for 2 h. The absorbance of each well at a wavelength of 480 nm was measured using a microplate reader (Bio-Rad, Hercules, CA, USA). Each experiment was repeated three times.

Flow cytometry

Cell apoptosis was detected using Annexin V-FITC/PI double staining. Cells were collected 48 h post-transfection, and the cell concentration was adjusted to 1 × 106 cells/mL, followed by fixing with pre-cooled 70% ethanol solution at 4 °C overnight. After being washed twice with PBS, 100 µL of the cell suspension (containing no fewer than 106 cells/mL) was taken and, after two further washes with PBS and centrifugation, cells were resuspended in 200 µL of binding buffer. 10 µL of Annexin V-FITC and 5 µL of PI were gently mixed in, and the mixture was left to react in the dark at room temperature for 15 min. After adding 300 µL of binding buffer, cell apoptosis was assessed using a flow cytometer (Attune NxT, Thermo Fisher, USA) with an excitation wavelength of 488 nm.

Transwell experiment

In vitro cell migration and invasion assays were conducted using 8 μm-pore sizes Transwell chambers from Corning Incorporated (USA) within a 24-well plate format. Within these 8 μm-pore polycarbonate membrane Transwell chambers, the lower chamber was pre-loaded with 600 mL of DMEM medium containing 20% FBS and allowed to equilibrate at 37 °C for 1 h. MCF-7 cells, post 48-hour transfection, were resuspended in FBS-free DMEM medium and seeded in the upper chamber at a density of 1 × 10^6/mL, followed by incubation at 37 °C, 5% CO2 for 24 h. Subsequently, the Transwell chambers were washed twice with PBS, each time for 5 min, then fixed with 4% paraformaldehyde for 20 min, followed by three washes with PBS, each for 5 min. Afterward, cells were stained with 0.1% crystal violet for 10 min and washed three times with PBS, each time for 5 min. Surface cells were removed using a cotton swab, and the cells that had migrated through the Transwell chambers were observed under an inverted fluorescence microscope (Nikon TE2000, Japan), photographed in 5 random fields of view, and counted. The average cell number that migrated through the Transwell chambers was recorded for each group. Each experiment was repeated three times.

Nude mice tumor formation and metastasis experiment

Six-week-old BALB/c nude mice (Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing, China) were caged and housed in an SPF-grade animal laboratory with a humidity of 60 − 65% and a temperature of 22–25 °C. They were provided with free access to food and water under a 12-hour light and dark cycle and allowed to adaptively feed for one week before the commencement of the experiment. The experimental procedures and animal usage plan have been approved by the Animal Ethics Committee.

A number of 2 × 105 MCF-7 cells/0.2 mL (including oe-NC, oe-CYP24A1, and oe-TFPI2 variations) were respectively injected subcutaneously into the left axilla of the nude mice, and prior to inoculation, the cells were premixed with Matrigel gel (Solebao YZ-356234-5 ml) to prepare the respective cell suspensions. The nude mice were randomly divided into 3 groups (6 per group): (1) oe-NC group: injected with HCC70 cells transfected with oe-NC; (2) oe-CYP24A1 group: injected with MCF-7 cells transfected with oe-CYP24A1; (3) oe-TFPI2 group: injected with MCF-7 cells transfected with oe-TFPI2. Tumor sizes were measured from day 5 post-injection, with measurements taken every 5 days to record tumor growth, and after 30 days, tumors from each group of nude mice were harvested for subsequent experiments.

For the metastasis nude mouse xenograft model, the mice were randomly assigned to 3 groups, each consisting of 6 animals. Six weeks after being raised, a breast cancer lung metastasis model was established by injecting cells (2 × 105 cells/0.2 mL) into the nude mice via tail vein injection. Upon completion of the breeding period, the mice were dissected via the cervical spine, and lung tissue was collected for H&E staining to assess lung metastasis.

H&E staining

Lung tissue samples from each group were first washed with saline, then fixed in 4% paraformaldehyde for 30–50 min, followed by processes of washing, dehydration, clarification, wax immersion, embedding, and sectioning. The tissue sections were flattened and adhered to glass slides, then dried in a 45 °C incubator, followed by deparaffinization and washing for 5 min with progressively diluting alcohol concentrations and distilled water. Hematoxylin staining was performed for 5 min, after which the sections were rinsed under tap water for 3 s, followed by differentiation in 1% hydrochloric alcohol for 3 s and eosin staining for approximately 3 min. Finally, the sections underwent dehydration, clarification, and cover-slipping processes. Tissue sections were observed under a microscope. A total of 18 nude mouse samples were used in this experiment.

Statistical analysis

Statistical data analysis in this study was conducted using SPSS 21.0 statistical software from IBM. Quantitative data were presented as mean ± standard deviation. Initially, data from cancer tissues and adjacent non-cancerous tissues were tested for normality and homogeneity of variance; if they met the requirements of normal distribution and homogeneity of variance, paired t-tests were used for comparison; unpaired t-tests were used for comparison between the other two groups; one-way analysis of variance (ANOVA) was used for comparison among multiple groups. For comparisons of data at different time points, cell activity was analyzed using two-way ANOVA, while tumor data were analyzed using variance analysis for repeated measures data. A P-value of less than 0.05 was considered to indicate a statistically significant difference.