Cel (purity>98%) was obtained from Shanghai Yuanye Bio-Technology Co., Ltd. (Shanghai, China). Cell counting kit-8 (CCK-8), Matrigel, paraformaldehyde (4%) and dimethyl sulfoxide (DMSO) were derived from LABLEAD Biotechnology Co., Ltd. (Beijing, China). Crystal violet was obtained from Beyotime Biotechnology (Shanghai, China). Annexin V-FITC/PI apoptosis detection kit was supplied by BD Biosciences (San Jose, CA, USA). Primary antibodies to p-EMT markers, EGFR, MEK, p-MEK, ERK, p-ERK were obtained from Proteintech Group, Inc. (Wuhan, China). Antibody to GAPDH was purchased from ABclonal Biotech Co., Ltd. (Wuhan, China). Antibody to p-EGFR was supplied by Abcam (Cambridge, UK).

Human TNBC cell lines MDA-MB-231 and BT-549 were obtained from Fenghui Biotechnology Co., Ltd. (Hunan, China) and Shanghai Fuheng Biotechnology Co., Ltd. (Shanghai, China) respectively. The MDA-MB-231 and BT-549 cells were incubated in complete Dulbecco’s modified Eagle’s medium (DMEM, Shanghai XP Biomed Ltd., China) and RMPI-1640 (Gibco, USA) supplemented with 1% penicillin-streptomycin (Gibco, USA) and 10% Fetal bovine serum (FBS, Corning, USA) respectively in an incubator with 5% CO₂ at 37 °C.

To construct TNBC-AR model, TNBC cells (5 × 10⁵) were seeded in Ultra-Low Attachment 6-well plates (Beyotime Biotechnology, Shanghai, China) for 48 h. The cells transferred to normal plates and re-adherent were considered as TNBC-AR cells. For model construction, cells at passages 3-4 were used. In order to obtain sufficient cell numbers for subsequent experiments, these cells were further expanded in culture. Therefore, the actual experiments were performed using cells at passages 6–8. Before modeling, no significant difference in proliferation rate was observed between the cloned cells and the parental cells; however, the clonogenic capacity declined after prolonged culture (notably beyond passage 11). To confirm the stability of the anoikis-resistant phenotype across passages, we have supplemented the anoikis assay data in Supplementary Figure S1, comparing cells at passage 4 (used for modeling) with those at passages 6 and 8 (used in experiments). The results indicate no significant difference in anoikis rates among passages 4, 6, and 8, supporting the phenotypic stability during the experimental window.

TNBC cells were seeded into 96-well plates (5000/well) and treated with various concentrations of Cel (0, 0.5, 1.0, 2.0, 4.0, 6.0, 8.0 μM). After culturing for 24h, 10 μL CCK-8 working solution was added. Then the 96-well plates were placed in an enzyme-labeled instrument to measure the absorbance of each well at 450 nm after incubating for 3 h. The absorbance data was recorded and calculated using GraphPad Prism 9.0.

Anoikis of TNBC cells was detected by using the Annexin V-FITC/PI apoptosis detection kit. Briefly, cells (3 × 10⁵) were seeded to Ultra-Low Attachment 6-well plates and treated with various concentrations of Cel or EGF for 24 h. Then cells were collected and stained with Annexin V-FITC and PI solution at 25 °C under dark conditions. The anoikis rate of cells was detected using flow cytometer (BD Biosciences, CA, USA).

TNBC cells (5 × 10⁵) were transferred into 6-well plates and then scratched using a 200 μL pipette once they achieved 80% confluence. Subsequently, FBS-free medium with different drugs was used to replace the primary medium. Fluorescent inverted microscope (magnification, ×100) was used to image the wound gap of 0, 24, or 48 h.

Invasion assay was detected using 24-well plates and 8.0 μm pore polycarbonate membrane inserts (Corning, USA) coated with 50 μL Matrigel. TNBC cells (5 × 10⁴) were plated on the upper chambers with FBS-free medium containing different concentrations of Cel or EGF. And the lower chambers were supplied with complete medium. After 24h, the TNBC cells that invaded into lower chambers were fixed using 4% paraformaldehyde and stained using crystal violet and eventually imaged and counted in 3 randomly regions using microscope (magnification, ×200).

TNBC cells (5 × 10⁵) were seeded in 6-well plates and treated with Cel for 24 h. Cell morphology was imaged using microscope.

TNBC cells (5 × 10³) were seeded in Ultra-Low Attachment 96-well plates (Beyotime Biotechnology, Shanghai, China) and treated with Cel for 48 h. Cell aggregation was imaged using microscope.

Colony formation assay was performed as previously described (Wang et al., 2023).

Differential genes regulated by Cel were obtained from SRA database (https://www.ncbi.nlm.nih.gov/sra/, PRJNA1426779). Target genes of TNBC metastasis were obtained from GeneCards database (https://www.genecards.org). EGFR expression profiles were downloaded from GENT2 database (http://gent2.appex.kr/gent2/). The volcano plot, Venn diagram, GO and KEGG analysis, Kaplan-Meier analysis, and EGFR expression of various BC subtypes were plotted using https://www.bioinformatics.com.cn (last accessed on 1 February 2025), an online platform for data analysis and visualization.

Western blotting was performed as previously described (Li et al., 2025).

GraphPad Prism 9.0 program was used for the statistical analysis. Statistical significances were carried out by unpaired t-tests or one-way ANOVA. P-values ≤0.05 were considered significant. (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). All data were shown as mean ± standard deviation.

An anoikis resistant cell model was established using the MDA-MB-231 and BT-549 cell lines as described previously. To investigate the effect of modeling, we used CCK-8 and flowcytometry to access the survival rates and anoikis rates of TNBC-AR cell lines. As shown in Figures 1A,B, the modeling TNBC-AR cells exhibited higher survival rates and lower anoikis rates in suspension conditions, indicating TNBC-AR cells are more resistant to anoikis. Then wound-healing and Transwell assays were conducted to observe the migration and invasion ability of TNBC-AR cells. As shown in Figures 1C,D, the MDA-MB-231-AR and BT-549-AR cells exhibited profound migration and invasion ability compared to the parental (P) cells. These observations indicated that the TNBC-AR cells which survive from ECM detachment are more easily to migrate and invade into distant tissues.

Cel, one of the key biologically active compounds in Tripterygium wilfordii Hook. F, shows a pentacyclic triterpene structure as in Figure 2A. A CCK-8 assay was performed to reveal the ability of Cel on suppressing the proliferation of MDA-MB-231-AR and BT-549-AR cells, and Cel at 1 and 2 μM inhibited TNBC-AR cells' viability by nearly 50% (Figure 2B). After calculation, it was obtained that the IC₅₀ values of MDA-MB-231-AR and BT-549-AR cells were 1.510 μM and 1.673 μM respectively. Subsequently, we selected concentrations (0.5, 1 and 2 μM) of Cel for further investigation. As shown in Figures 2C,D, Cel inhibited the ability of MDA-MB-231-AR and BT-549-AR cells to form colonies. Besides, treatment with Cel led to lower cell density and more apoptotic morphology changes in MDA-MB-231-AR and BT-549-AR cells (Figure 2E). These results reveal that Cel significantly suppresses the proliferation activity of TNBC-AR cells.

To investigate the potential impact of Cel on the invasivity and migratory activity of MDA-MB-231-AR and BT-549-AR cells, Transwell assay and wound-healing assay were conducted. As shown in Figure 3A, Cel can suppress the invasive ability of MDA-MB-231-AR and BT-549-AR cells in a dose-dependent manner. Similarly, treatment of Cel dose dependently leads to the suppression of migratory activity of MDA-MB-231-AR and BT-549-AR cells (Figures 3B,C). These results suggest that Cel could inhibit the high metastatic potential of TNBC-AR cells.

Collective migration is a hallmark of cancer metastasis, especially in haematogenous routes (Alexandrova et al., 2025). Compared to single CTC, formation of CTC clusters can provide survival advantage for tumor cells within vascular system (Ildiz et al., 2023). Then we further examined the effect of Cel on the aggregation and cluster formation of TNBC-AR cells in suspension conditions (Figure 4A). As shown in Figure 4B, Cel significantly inhibited the aggregation and formation of CTC clusters in MDA-MB-231-AR and BT-549-AR cells. In addition, Cel enhanced the anoikis of MDA-MB-231-AR and BT-549-AR cells (Figure 4C). And in the treatment of low dose, MDA-MB-231-AR cells were more resistant to anoikis than BT-549-AR cells. These results suggest that Cel could inhibit aggregation and cluster formation of TNBC-AR cells and promote the apoptosis of TNBC-AR cells during the process of metastasis.

After proving the effects of Cel on TNBC-AR cells metastasis, the signaling pathways and molecular mechanism underlying these effects were explored by transcriptome analysis. The DEG analysis revealed that 419 genes were significantly downregulated, and 234 genes were significantly upregulated after Cel treatment in TNBC cells (Figure 5A). As shown in Figure 5B, 186 common targets between Cel-associated targets and TNBC metastasis-related targets were identified via using GeneCards database. Subsequently, we performed GO and KEGG enrichment analysis of the common targets and the top 10 enriched results are illustrated in Figures 5C,D. Notably, cytokine-cytokine receptor interaction emerged prominently in the KEGG analysis, which includes the EGFR signaling pathway, a signaling pathway that play a crucial role in the anoikis resistance of tumor cells and in the metastasis of TNBC. These results suggest that EGFR signaling pathway may be a significant pathway suppressed by Cel in the process of TNBC metastases.

To study the underlying clinic relevance of our analysis, we investigated the association of high EGFR expression and BC-specific survival probability, and the expression of EGFR in various subtypes of BC. As shown in Figures 6A,B, BC patients with EGFR high expression exhibits an obvious lower survival probability, and the TNBC subtype exhibits the highest expression of EGFR, which may explain the high mortality of TNBC patients. Subsequently, the effect of Cel on the EGFR signaling pathway was validated in MDA-MB-231-AR cells. As shown in Figure 6C, Cel apparently inhibited the phosphorylation of EGFR (p-EGFR) on the protein level. Furthermore, the downstream molecules of EGFR including p-MEK and p-ERK were significantly reduced after Cel treatment. Besides, E-cadherin and N-cadherin, two biomarkers of p-EMT were both inhibited upon Cel treatment, indicating the suppression of p-EMT state of MDA-MB-231-AR cells (Figure 6D). Collectively, these results reveal that Cel may potentially inhibit metastatic traits by interfering with the EGFR/MEK/ERK pathway and reduce the expression of E-cadherin and N-cadherin.

To further validate the role of EGFR/MEK/ERK signaling pathway in the anti-metastasis effects of Cel treatment, we employed EGFR specific activator EGF protein combined with Cel to treat MDA-MB-231-AR cells in the following study. As shown in Figure 7A, protein levels of EGF and p-EGFR were increased after EGF treatment, and the inhibitory effects of Cel were reversed by EGF, so as the downstream proteins of EGFR. In addition, the Cel-inhibited expression of E-cadherin and N-cadherin was also reversed by EGF, indicating that Cel could impact the p-EMT state via EGFR/MEK/ERK signaling pathway. Moreover, EGF treatment can diminish inhibited invasion and migration ability of MDA-MB-231-AR cells with Cel treatment (Figures 7B,C). Then the effects of EGF on cluster formation and anoikis rate of MDA-MB-231-AR cells were evaluated. As shown in Figures 7D–F, the combined treatment of EGF and Cel reversed the suppressed cluster formation and enhanced anoikis rate compared with only Cel treated group. Taken together, these results indicate that the EGFR/MEK/ERK signaling pathway is profoundly involved in the anti-metastasis effects of Cel.