The Cancer Genome Atlas (TCGA-BRCA) contains the data of 1109 BC and 113 non-carcinoma breast samples. The Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) has the data of 1904 BC samples, whereas the Gene Expression Omnibus (GEO) database, specifically the GSE96058 dataset, includes the data of 3409 BC samples. The Molecular Signatures Database (MSigDB) v7.4 was adopted in this study to identify the fatty acid metabolism genes, including those related to the KEGG fatty acid metabolism pathway, Reactome fatty acid metabolism, and hallmarks of fatty acid metabolism [10, 11]. The overlapping genes were removed from these identified genes, and 309 fatty acid metabolism-related genes (FMGs) remained. Subsequently, these 309 FMGs were intersected with the total genes from the TCGA-BRCA, METABRIC, and GSE96058 datasets, and 269 reliable FMGs were identified and used for further analysis.

Weighted gene coexpression network analysis (WGCNA) is a systems biology method used to describe the patterns of gene associations across different samples [12]. Using the imageGP website (https://www.bic.ac.cn), a WGCNA was conducted in this study for the expression data of the above-identified 269 FMG genes in cancer and tumor tissues from the TCGA-BRCA dataset, and the gene adjacency heatmaps were obtained. Next, univariate Cox analysis was conducted to identify the correlations of the genes in the module most associated with tumors to the overall survival rate in the TCGA-BRCA cohort, with an objective of identifying the hub genes related to fatty acid metabolism (p 0.05 indicated statistical significance).

Thereafter, the DECR1 protein levels in tumor and matched non-carcinoma tissues were determined from the Human Protein Atlas (HPA) database (https://www.proteinatlas.org/). Immunohistochemical (IHC) staining was semi-quantitatively scored by multiplying staining intensity (0–3) with the percentage of positive cells (1–4), yielding a total score ranging from 0 to 12. Intensity was graded as: 0 = negative, 1 = weak, 2 = moderate, 3 = strong. Proportion scores were: 1 = $\le$25%, 2 = 26–50%, 3 = 51–75%, 4 = $>$75%. The percentage of positively stained cells was also scored on a 4-point scale: 1 = $\le$25%, 2 = 26–50%, 3 = 51–75%, 4 = $>$75%. The final IHC score was calculated by multiplying the intensity score by the proportion score, yielding a composite score ranging from 0 to 12. Furthermore, the relationship between DECR1 and overall survival in patients with breast cancer was investigated using the Kaplan-Meier Plotter database (https://kmplot.com/analysis/). The Tumor and Immune System Interaction Database (TISIDB) database [13] was used to analyze the relationships of DECR1 with breast cancer immune subtypes and breast cancer subtypes. The immune subtypes were identified as C1–C6 according to the criteria reported by Thorsson et al. [14].

The natural product library of TargetMol (TargetMol, Boston, MA, USA), which contains the data of about 19,000 natural small-molecule compounds, was used for drug screening. All compounds were docked to the binding port defined by the DECR1 protein (ID: 1W6U) using a three-step method. High-throughput screening was completed in step 1, retaining 50% of the compounds; standard precision was performed in step 2, and 50% of the compounds from the previous step were retained; extra precision was conducted in step 3, in which, 50% of the compounds from step 2 were retained. By setting the compound properties to 0 H bond donor 6 and 150 Mol weight 725, the compounds were further screened, revealing 1258 compounds after screening.

These 1258 compounds were zipped and input into Canvas to calculate the binary fingerprints of all small molecules. Then, hierarchical clustering was performed on the Tanimoto similarity matrix. From each class, the docking score (in kcal/mol, where a smaller value represents the highest affinity between the receptor and ligand) was selected, and the highest number of small-molecule compounds was retained. Finally, 100 small molecule compounds belonging to different chemical structures were obtained. An intersection analysis between these 100 obtained small molecule compounds and the drugs already listed in the Food and Drug Administration (FDA) database was performed, revealing two drugs: atorvastatin calcium and pravastatin sodium. The ligands pravastatin and atorvastatin, along with the DECR1 protein (1W6U), were subsequently submitted to SwissDock (http://www.swissdock.ch) for molecular docking using AutoDock Vina (The Scripps Research Institute, La Jolla, CA, USA) [15, 16]. Finally, Proteins Plus (https://proteins.plus/) was used for visualization and processing, and images were obtained.

This experiment was conducted in accordance with the standard Surface Plasmon Resonance (SPR)-related procedures. The experimental design and parameters are shown in the table below. In brief, the carboxyl (COOH) chip was installed on the OpenSPRTM instrument (Nicoya Lifesciences, Kitchener, ON, Canada), the sample ring was cleaned with PBS buffer, and the chip was activated with 1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide/N-Hydroxysuccinimide (EDC/NHS, 1:1) solution when the signal reached the baseline. The DECR1 ligand (200 µL) was loaded followed by incubation for 4 min. The analyte was diluted with buffer (PBS + 1% DMSO) and loaded at a rate of 20 µL/min. The protein and ligand binding time were 240 s, and natural dissociation was allowed for 300 s. Data analysis was performed using the one-to-one analysis model in TraceDrawer (Ridgeview Instruments AB, Uppsala, Sweden) software.

The BC cell lines BT474, MDA-MB-231, and 4T1 were purchased from ATCC (American Type Culture Collection, Manassas, VA, USA). BT474 and MDA-MB-231 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (SH30022.01B, HyClone, Logan, UT, USA) supplemented with 10% fetal bovine serum (SH30088.03, HyClone, Logan, UT, USA), 100 U/mL penicillin, and 100 µg/mL streptomycin (C0222, Beyotime, Suzhou, Jiangsu, China), and 4T1 cells were cultured in RPMI-1640 (SH30809.01B, HyClone, Logan, UT, USA) complete medium. All cells were incubated at 37 °C in 5% CO₂ for culture, and the medium was changed every 2–3 days. All cell lines used in this study were validated by Short Tandem Repeat (STR) profiling and tested negative for mycoplasma contamination.

The logarithmic BC cells were seeded and cultured in the wells of 96-well plates overnight, followed by further seeding and culture in the medium containing atorvastatin calcium (T3116, TargetMol, Boston, MA, USA) or pravastatin sodium (T0672, TargetMol, Boston, MA, USA) for 48 h at 37 °C with 5% CO₂. The cell seeding concentrations used were 0, 3.125, 6.25, 12.5, 25, 50, 100, and 200 µM. After 48 h of cell culture, the breast cancer cells were treated with the CCK-8 solution (BS350B, Biosharp, Hefei, Anhui, China), and absorbance values were read at 450 nm.

DMEM (with 10% FBS) and breast cancer cells were added to the top Transwell chamber (3464, Corning, Corning, NY, USA), and only DMEM (with 20% FBS) was added to the bottom chamber. At 48 h of incubation at 37 °C, 0.1% crystal violet was added for cell staining, followed by microscopic observation and cell counting.

The cells from the different treatment groups (1500 rpm, 5 min) were centrifuged, and the cells were collected in centrifuge tubes. The samples were fixed with 3% pentanediol and then with 1% osmium tetroxide, followed by dehydration with gradient acetone and embedding in the Epon-812 embedding agent. An ultrathin slicing mechanism (Leica Microsystems, Wetzlar, Hesse, Germany) was used to prepare ultrathin slices of about 60–90 nm, and a JEM-1400FLASH transmission electron microscope (JEOL Ltd., Tokyo, Japan) was used to acquire images of these slices when observed.

A DCFH-DA probe (Beyotime, China, S0033M) was used to evaluate the intracellular ROS levels. A FerroOrange probe (F374, Dojindo Laboratories, Kumamoto, Japan) was used to detect the intracellular Fe²⁺ content. A Liperfluo assay kit (DOJINDO, Japan, L248) was used to measure the levels of lipid peroxides. Each experiment was performed following specific protocols, and a fluorescence microscope (Leica Microsystems, Wetzlar, Hesse, Germany) was used for observation.

The total RNA in the samples was extracted with TRIzol Reagent (15596026, Invitrogen, Carlsbad, CA, USA) and then reverse transcribed into cDNA using the HiScript II Q RT SuperMix for qPCR Reagent Kit (R223-01, Vazyme, Nanjing, Jiangsu, China). The RT-PCR system was configured using ChamQ Universal SYBR qPCR Master Mix (Q711-02, Vazyme, Nanjing, Jiangsu, China), and the primer sequences used for qPCR are listed in Table 1. The amplification reactions were performed using a real-time qPCR instrument (Archimed-X6, ROCGENE, Wuhan, Hubei, China). Gene expression was calculated using the 2^{-Δ⁢Δ⁢ct} approach, with GAPDH used as an endogenous reference. The primers used were as in Table 1.

Radioimmunoprecipitation assay (RIPA) lysis buffer supplemented with a protease inhibitor cocktail was used for cell lysis. A BCA kit (P0012, Beyotime, Suzhou, Jiangsu, China) was used to determine the protein content. Protein aliquots were then subjected to western blotting. The following primary antibodies were used: anti-$\beta$-actin (ab6276, Abcam, Cambridge, Cambridgeshire, UK), anti-DECR1 (ab198848, Abcam, Cambridge, Cambridgeshire, UK), and anti-ACSL4 (sc-365230, Santa Cruz Biotechnology, Dallas, TX, USA). HRP-conjugated secondary antibodies (SA00001-1 and SA00001-2, Proteintech, Chicago, IL, USA) were utilized to detect the bound antibodies using an enhanced chemiluminescence (ECL) kit (P0018S, Beyotime, Suzhou, Jiangsu, China).

Female BALB/c mice (about 6 weeks old) were procured from Hunan SJA Laboratory Animal Co., Ltd. (Changsha, Hunan, China). The experimental animals were raised at the experimental animal center of Southwest University under specific pathogen-free conditions. The purchase, shipping, housing, care, and euthanization of all mice were in accordance with the guidelines of the Institutional Animal Care and Use Committee of Southwest University.

The 100 µL of a 4T1 breast cancer cell suspension with 1 $\times$ 10⁶ cells were injected subcutaneously into the axillary region of BALB/c mice to construct the subcutaneous breast cancer xenograft model. At a volume of about 50 mm³, atorvastatin calcium (20 mg/kg and 40 mg/kg) were administered intraperitoneally, once every 2 days, for 6 times. The tumor volume was then calculated as follows: volume = (length $\times$ width²)/2 at 2-day intervals where length refers to the longest dimension and width refers to the perpendicular shorter axis. To ensure accuracy, all measurements were performed using digital calipers by trained personnel in a blinded manner, and tumors with irregular morphology or poorly defined boundaries were excluded from volumetric analysis. After receiving six doses of the treatment, the mice were euthanized for further analysis and the required tissue samples were retrieved. All animals were euthanized humanely using carbon dioxide (CO₂) inhalation, with a fill rate of 20% chamber volume per minute, in accordance with the AVMA Guidelines (2020) and approved by the Institutional Animal Care and Use Committee (IACUC).

Mouse tumor, liver, kidney, and spleen samples were collected and immersed in 4% paraformaldehyde at ambient temperature. The paraffin-embedded samples were later sliced into 5 µm sections for hematoxylin-eosin (H&E) staining.

After 24 h of immersion in 10% neutral formalin, the tumor tissues were subjected to paraffin embedding and then sliced into 5 µm sections. Afterward, the sections were deparaffinized and rehydrated, followed by antigen retrieval by boiling in citrate buffer (pH 6.0) for 15 min. Thereafter, 3% hydrogen peroxide was added to block the endogenous peroxide for 10 min. The sections were blocked with 5% goat serum for 1 h and then incubated overnight at 4 °C with DECR1 antibody (Abcam, England, ab198848) and Ki67 antibody (Abcam, England, ab15580). The sections were then washed followed by 1 h of incubation with an HRP-labeled secondary antibody and then 3,3^′-Diaminobenzidine (DAB) staining for 5 min. Hematoxylin was used for counterstaining, and the slides were dehydrated, cleared, and mounted. Images of the sections were captured using a microscope (Leica Microsystems, Wetzlar, Hesse, Germany).

The results were presented as means $\pm$ standard deviations (SDs). Intra-group differences were compared using the Student’s t-test, whereas inter-group differences were analyzed using one-way ANOVA. Every assay was conducted at least three times. p 0.05 indicated statistical significance. GraphPad Prism 9.0 software (GraphPad Software, San Diego, CA, USA) was used for statistical analysis.

The effects of the reprogramming of fatty acid metabolism on BC occurrence and development are undeniable. Therefore, the genes associated with fatty acid metabolism were screened to identify tumor therapeutic targets. The fatty acid metabolism-related genes associated with BC were identified from the TCGA-BRCA (n = 1109), METABRIC (n = 1904), and GSE96058 (n = 3409) datasets, and the intersection revealed 269 potential genes (Supplementary Table 1, Fig. 1A). Subsequently, WGCNA was performed for these 269 genes to identify the module most strongly associated with tumors. WGCNA revealed a positive correlation between the brown module and the tumor phenotype (R = 0.334, p 0.0001) (Fig. 1B,C), and 50 genes from this module were selected for subsequent analysis (Supplementary Table 2) using a univariate Cox analysis. DECR1 was consequently identified as a risk factor for breast cancer, whereas PTGS2, ALDH3A1, PLA2G4A, ALDH2, ACSL5, and CYP1B1 were identified as protective factors (Fig. 1D). Subsequent analysis focused on determining the role of DECR1 in breast cancer.

Fig. 1.

DECR1 has potential as a candidate anti-BC target. (A) Identification of 269 reliable genes related to fatty acid metabolism, with a Venn diagram showing these 269 overlapping FMGs from the TCGA-BRCA, METABRIC, and GSE96058 datasets. (B) Characteristic gene adjacency heatmaps of the correlations between normal breast and tumor tissue module genes. (C) Correlations of gene significance with the module membership of the brown module with the highest absolute R-value. (D) Univariate Cox analysis conducted for screening the prognostic genes in the brown module. (E) Representative IHC staining of DECR1 in normal and tumor breast tissues from the Human Protein Atlas (HPA), along with semi-quantitative scoring of staining intensity. (F) Survival analysis for determining the correlation level of DECR1 expression in relation to patient outcomes (https://kmplot.com/). (G,H) TISIDB website findings showing significant relationships of DECR1 with immune subtypes and breast cancer subtypes. C1 to C6 denote the different immune subtypes. ** p 0.01. FMGs, fatty acid metabolism-related genes; TCGA-BRCA, the Cancer Genome Atlas Breast Invasive Carcinoma dataset; METABRIC, Molecular Taxonomy of Breast Cancer International Consortium dataset; IHC, Immunohistochemistry staining; DECR1, 2,4-Dienoyl-CoA Reductase 1.

Compared to its levels in the adjacent tissues, DECR1 was significantly overexpressed in breast cancer tissues (Fig. 1E). Next, based on the average expression of DECR1 in TCGA-derived breast cancer tissues, the patients were classified into DECR1 low-expression or DECR1 high-expression groups using a cutoff of 50%. DECR1 exhibited significant differential expression in BRCA (Kruskal-Wallis test, p = 9.46 $\times$ 10^-7), potentially associated with cluster groups (C1–C6) or distinct molecular subtypes (e.g., Basal-like, HER2, Luminal) (Fig. 1G,H). According to the survival analysis, high DECR1 expression was negatively correlated to patient prognosis (Fig. 1F), indicating that high DECR1 expression was associated with poor outcomes. These results suggested that DECR1 is closely associated with BC prognosis, although the underlying mechanisms remain unclear.

The preliminary research showed that DECR1 could serve as a prognostic indicator with a negative correlation to patient prognosis. Therefore, searching for potential candidate drugs targeting DECR1 has clinical significance and prospects. Accordingly, computer-aided screening was conducted in this, for which the ligand NADP in the DECR1 complex structure (Fig. 2A) was defined as the binding pocket and the top 100 compounds screened were intersected with the FDA-marketed drug database. As a result, atorvastatin calcium (AC) and pravastatin sodium (Fig. 2B,C), which could bind to the DECR1 protein, were identified. AC formed hydrogen bonds with the DECR1 protein at Thr69, Gly72, Ala146, Phe249, and Asn144. Additionally, the Phe249 and Leu71 residues were involved in hydrophobic interactions (Fig. 2D). Pravastatin sodium formed hydrogen bonds at Tyr199, Ile243, Thr245, Gly147, and Ser210, while hydrophobic interactions were noted with Thr197 and Leu71 (Fig. 2E). To investigate the direct interaction between DECR1 and atorvastatin calcium (AC), surface plasmon resonance (SPR) analysis was performed (Fig. 2F). A concentration-dependent binding curve demonstrated that AC binds to DECR1 with high affinity and specificity in vitro.

Fig. 2.

Binding of DECR1 to pravastatin sodium and atorvastatin calcium. (A) The 3D structure of DECR1 (PDB ID: 1W6U). (B) Chemical structure of pravastatin sodium. (C) Chemical structure of atorvastatin. (D) Schematic for the molecular docking between pravastatin sodium and DECR1. (E) Molecular docking between atorvastatin calcium and DECR1. (F) Concentration gradient binding curve for the binding of DECR1 and atorvastatin calcium by SPR experiments and we confirmed the high affinity and specificity of AC binding to DECR1 in vitro. AC, atorvastatin calcium.

Furthermore, the effects of AC and PS on BC cell proliferation were evaluated, and it was confirmed that AC could suppress MDA-MB-231 and BT474 cell growth, with IC50 values of 39.70 µM and 7.74 µM, respectively (Fig. 3A,B). However, BC cells are insensitive to different concentrations of pravastatin sodium; therefore, it was speculated that AC has greater potential than pravastatin sodium to efficiently treat BC. Therefore, further research was conducted using AC. BC cell migration ability was markedly inhibited following AC treatment (Fig. 3C,D). The in vivo research revealed that AC administration significantly inhibited tumor development (Fig. 4A,B), and a preliminary in vivo safety assessment revealed that AC caused little harm to key organs in the tumor burden model, as evidenced by HE staining showing no significant morphological changes in the liver, kidney, or spleen. Additionally, immunohistochemical staining of tumor tissues demonstrated a marked decrease in Ki67 and DECR1 expression after AC treatment, indicating suppressed tumor cell proliferation and downregulation of DECR1 in vivo (Fig. 4C,D). In summary, our findings suggest that the antitumor effect of AC may be linked to its regulation of DECR1.

Fig. 3.

Atorvastatin calcium (AC) has antitumor effects. (A,B) Compared to pravastatin sodium, AC markedly suppressed MDA-MB-231 (A) and BT474 (B) cell proliferation, with IC₅₀ values of 39.70 µM and 7.74 µM, respectively. (C,D) AC treatment inhibits cell migration in breast cancer cells. (C) Representative images of migrated MDA-MB-231 and BT474 cells treated with AC at 1 µM and 5 µM, as assessed by Transwell migration assay, scale bar = 249 µm. (D) Quantification of migrated cells from (C). ** p 0.01.

Fig. 4.

The in vivo antitumor effects of AC. (A) Intraperitoneal administration of AC (20 mg/kg or 40 mg/kg) significantly inhibited tumor growth in the 4T1 subcutaneous tumor model. (B) The tumor growth curves for different treatment groups. (C) Mouse body weight changes during drug treatment. (D) HE staining of tumor, liver, spleen, and kidney tissues (scale bar = 498 µm) showed no significant morphological changes in the liver, kidney, or spleen following AC treatment. Immunohistochemical staining of Ki67 and DECR1 in tumor tissues (scale bar = 124.5 µm) revealed a marked decrease in expression, indicating suppressed cell proliferation and DECR1 expression after treatment. (E,F) Transmission electron microscopy revealed abnormal inner mitochondrial membrane structures in AC-treated MDA-MB-231 (E) and BT474 (F) cells, which are the typical features of ferroptosis. In each panel, the left image shows the overall cellular morphology (scale bar = 2 µm), while the right image displays a magnified view of the yellow-boxed region, highlighting mitochondrial damage (scale bar = 1 µm). Red arrows indicate condensed and ruptured mitochondrial cristae, characteristic of ferroptotic morphology. ** p 0.01. HE, hematoxylin-eosin.

Clinically, AC is used mainly to treat hyperlipidemia, atherosclerosis, and other lipid metabolism disorders. Accordingly, combined with the findings of this study, it was speculated that the potential role of AC is to affect fat metabolism in cells. Statins have been suggested to induce ferroptosis in different solid cancers by inhibiting the mevalonate pathway to reduce the biosynthesis of GPX4. Therefore, whether AC could induce ferroptosis in breast cancer cells was evaluated next in this study. Transmission electron microscopy revealed that AC treatment caused significant mitochondrial shrinkage, elevated mitochondrial membrane density, and decreased number of or no cristae (Fig. 4E,F).

Intracellular lipid peroxides are important factors involved in the induction of ferroptosis. The detection of lipid peroxidation reflects the degree of ferroptosis in cells. In this study, markedly increased lipid peroxide content was noted in the AC treatment group (Fig. 5A). The accumulation of ROS (Fig. 5B) and Fe²⁺ (Fig. 5C) is a crucial factor in the occurrence of ferroptosis. This study revealed that the ROS and Fe²⁺ levels were significantly increased after AC treatment (Fig. 5D,E). Furthermore, ferroptosis inhibitors (ferrostatins) could effectively rescue tumor cells from death. In contrast, the combination of a ferroptosis inducer (erastin) with AC resulted in a synergistic effect that led to promoted breast cancer cell death (Fig. 6A,B). These results indicated that AC promotes BC cell ferroptosis.

Fig. 5.

Detection of ferroptosis-related markers in breast cancer cells after AC treatment. (A) Immunofluorescence analysis results revealed markedly elevated lipid peroxidation levels in AC-treated breast cancer cells compared to those in control cells. (B) ROS levels were significantly elevated following AC treatment. (C) The intracellular Fe²⁺ content also markedly increased after AC treatment. (D,E) Relative fluorescence intensity (normalized to control) in MDA-MB-231 (D) and BT474 cells (E). Fluorescence intensity values were normalized to the control group. The scale bars represent 249 µm in all images. ** p 0.01, ns, not significant.

Fig. 6.

Investigation of related mechanisms. (A,B) Cotreatment of MDA-MB-231 (A) and BT474 (B) cells with ferroptosis inducers (erastin, 10 µM), ferroptosis inhibitors (ferrostatins, 10 µM), and AC for 48 h. (C,D) qRT-PCR analysis of ferroptosis-related gene expression in MDA-MB-231 (C) and BT474 (D) cells treated with AC (5 µM) for 24 h. Notably, ACSL4 mRNA expression was significantly upregulated in both cell lines (p 0.01), with a ~7.5-fold increase in MDA-MB-231 cells and a ~2.5-fold increase in BT474 cells. Expression levels of GPX4 and SLC7A11 showed moderate downregulation but did not reach statistical significance. Error bars represent SEM from three independent experiments. (E) Western blot analysis of DECR1 and ACSL4 protein expression in control and AC-treated MDA-MB-231 and BT474 cells. (F) Quantification of protein expression from (E), normalized to $\beta$-actin. AC treatment significantly decreased DECR1 expression and increased ACSL4 expression in both cell lines. ** p 0.01, ns, not significant. AC, atorvastatin calcium; DECR1, 2,4-Dienoyl-CoA Reductase 1; ACSL4, Acyl-CoA Synthetase Long-Chain Family Member 4.

In order to explore the specific molecular mechanisms underlying AC-induced ferroptosis in breast cancer cells, the expressions of ferroptosis-related genes were determined in this study. Compared to those in the control group, the expressions of GPX4 and SLC7A11 in the AC treatment group were markedly decreased (Fig. 6C,D), which was consistent with previously reported findings [17, 18]. Additionally, the expression of ACSL4 was obviously upregulated in this study. At the protein level, the expression of ACSL4 was negatively correlated to that of DECR1 (Fig. 6E,F). AC inhibits DECR1 expression but promotes ACSL4 expression. Therefore, it was understood that AC not only induces ferroptosis by reducing the biosynthesis of GPX4 but may also promote ferroptosis by activating ACSL4 expression.