2.1 Elevated CAPN2 Expression Predicts Adverse Clinical Outcomes and Enhanced Metastatic Potential in TNBC

Calpain 2 expression has been closely associated with an unfavorable prognosis in basal-like and TNBC [18]. To substantiate this correlation, we conducted a database analysis prior to studying the underlying molecular mechanisms. Among basal-like breast cancer patients with high gene expression of CAPN2 (encoding calpain 2), there was a tendency toward lower OS, distant metastasis-free survival (DMFS), and relapse-free survival (RFS) (Figure 1A). In addition, the forest plot of CAPN2 univariate Cox analysis by molecular subtype demonstrates that CAPN2 expression increases the hazard ratio (HR) in basal-like breast cancer (Figure 1B). Unlike in basal-like breast cancer, CAPN2 expression showed minimal impact on the risk of other subtypes of breast cancer, highlighting the potential significance of calpain 2 as a critical target only in TNBC (Figure S1A). In addition, upregulation of CAPN1 (encoding calpain 1), another major isoform of calpain, did not affect any type of survival (Figure S1B), underscoring the isoform-specific role of calpain 2 in influencing the cancer prognosis of TNBC. On the tissue microarray (TMA) slide containing TNBC biopsies, we observed a correlation between higher calpain 2 expression and tumor stage (Figure 1C). In TNBC patient tissues, elevated calpain 2 expression tended to align with higher tumor grades. Given that aggressiveness and metastasis are key factors in breast cancer prognosis [19], the increased risk and decreased survival may be associated with the mesenchymal phenotype. A heatmap analysis further revealed relationships between the expression of CAPN2 and mesenchymal and epithelial markers (Figure 1D,E). Cell lines categorized as the mesenchymal subtype of TNBC, basal B type, showed a tendency toward higher levels of mesenchymal markers and lower levels of epithelial markers, accompanied by higher CAPN2 and lower CAPN1 expressions. Moreover, the expression of CAPN2 in TNBC (IHC) and basal-like (PAM50) tissues exhibited a positive correlation with mesenchymal genes, such as FN1, ACTA2, and VIM (Figures 1F and S2A). Notably, the correlation between CAPN2 and mesenchymal genes in basal-like breast cancer was more pronounced than in other subtypes of breast cancer (Figure S2B). These findings suggest that the expression of CAPN2 but not CAPN1 may potentially contribute to an increased risk of TNBC and basal-like breast cancer characterized by enhanced mesenchymal gene expression.

2.2 CAPN2 Downregulation Inhibited Cancer Cell Proliferation, Migration, Invasion, and the EMT

We aimed to elucidate the isoform-specific roles of calpain 2 in the growth and metastasis of basal B or mesenchymal TNBC cells. Mesenchymal TNBC cell lines were selected and subjected to stable knockdown of CAPN2 or CAPN1 for further investigation (Figures 3 and S4). Notably, downregulating calpain 2, not calpain 1, significantly decreased both anchorage-dependent (Figures 2A–C and S4A–C) and anchorage-independent cellular growth (Figures 2D,E and S4D). Given that anchorage-independent growth of cancer cells is strongly associated with invasiveness and metastasis [20], the isoform-specific role of calpain 2 in cancer cell growth implies a potential connection between calpain 2 and the metastatic phenotype.

In addition to changes in cellular growth rate, cells with reduced calpain 2 expression exhibited morphological differences (Figure 2F). With CAPN2 knockdown, cells displayed a cobblestone-like morphology characterized by tight cell-cell adhesion, while cells transduced with an empty vector retained their spindle-like shape with a scattered distribution in culture. Calpain 2 downregulation significantly impaired not only cell migration (Figures 2G and S4E) but also invasion (Figure 2H), whereas calpain 1 downregulation showed little impact on these processes. These findings, along with the observed morphological differences and lesser growth upon CAPN2 knockdown, support the previous clinical studies and database analysis suggesting that calpain 2 is closely related to cancer cell metastatic prognosis in TNBC, but calpain 1 is not.

Considering the observed effects on migration and invasion, we explored whether calpain 2 regulates the epithelial–mesenchymal transition (EMT). Gene set enrichment analysis (GSEA) of TNBC patients indicated the enrichment of EMT-related gene sets in patients with high CAPN2 levels (Figures 3A,B and S5A). Downregulation of calpain 2 resulted in reversal of the EMT, characterized by an increase in epithelial markers and a decrease in mesenchymal markers (Figures 3C,D and S5B). We also developed CAPN2 overexpressing cellular models using basal A or epithelial TNBC cell lines and assessed the involvement of its enzymatic activity by overexpressing either wildtype CAPN2 or a mutant form (C105A) (Figure S3). Overexpression of calpain 2 downregulated E-cadherin expression, while an inactive mutant form of the enzyme did not decrease the E-cadherin level (Figure 3E). These changes in EMT-related protein markers were confirmed at the mRNA level (Figure 3F,G).

To identify the molecular mechanisms governing these changes induced by calpain 2 downregulation, we examined the levels of EMT-related transcription factors. Among these factors, the protein level of Twist consistently showed reductions (Figures 3H and S5C). Twist is a pivotal transcription factor governing genes associated with EMT [21-23], and its protein levels are regulated not only by transcriptional control but also by modulation of stability via MAPK and AKT signaling pathways [24]. Although phosphorylated MAPK and AKT remained unaffected by calpain 2 downregulation, the mRNA level of TWIST1 was significantly altered by CAPN2 regulation (Figures S6A and 3I,J). Correspondingly, we found a positive correlation between CAPN2 and TWIST1 transcripts (Figure S7B). Moreover, gene signature analysis revealed that an elevated mean expression of the gene set consisting of CAPN2 and TWIST1 in metastatic TNBC tissues was notably more significant than the individual gene expressions (Figure S6C–E). Overall, these findings suggest that the expression of calpain 2 impacts the EMT by regulating Twist expression at the transcriptional level in mesenchymal TNBC cell lines.

2.3 CAPN2 Downregulation Attenuates HIF1α Nuclear Localization and Limits TWIST1 Expression in TNBC

The observed decrease in TWIST1 expression at the mRNA level (Figure 3I,J) prompted us to investigate potential transcriptional regulation of TWIST1. Through an analysis of the TWIST1 promoter, we identified a set of candidate transcription factors (Table S3). Subsequently, we performed GSEA to uncover pathways associated with calpain 2 expression. In patients exhibiting elevated CAPN2 levels, GSEA revealed a significant correlation with gene expression signatures characteristic of hypoxia (Figures 3, 4, and S5A). HIF1α, a well-known transcription factor activated by hypoxic stimuli, regulates diverse gene expressions [25]. TWIST1 is one of the target genes regulated by HIF1α through direct binding to the hypoxia-response element (HRE) in the TWIST1 promoter [26]. It is known that the functionality of HIF1α as a transcription factor is influenced by its protein stability and subcellular localization [27]. Prior to the evaluation of HIF1α involvement in the regulation of TWIST1 expression, the role of HIF1α in EMT regulation was confirmed (Figure 4B). CoCl₂ treatment, known to stabilize HIF1α and enhance its activity [28], induces EMT by downregulating epithelial markers and upregulating mesenchymal markers. We subsequently assessed the nuclear localization of HIF1α and its impact on TWIST1 expression under CoCl₂-treated condition. Calpain 2 knockdown significantly reduced the nuclear proportion of HIF1α compared to control cells (Figures 4C and S7A). This reduction was further supported by RNA-seq analysis, which demonstrated calpain 2-dependent gene set enrichment in the hypoxia hallmark set (data not shown). CoCl₂ treatment induced nuclear accumulation of HIF1α and upregulated TWIST1 expression, which was attenuated by calpain 2 downregulation (Figure 4D). The chromatin immunoprecipitation (ChIP) assay revealed that calpain 2 knockdown resulted in the loss of HIF1α binding to the TWIST1 promoter (Figure 4E). These results indicate that the reduction of Twist in calpain 2 knockdown cells is attributed to alteration in the subcellular localization of HIF1α.

Filamin A cleavage has been reported as a regulator of HIF1α subcellular localization [29]. Therefore, we sought to determine whether the C-terminal fragment of filamin A contributes to the nuclear localization of HIF1α. In control and CAPN1 knockdown cells, significant cleavage of filamin A was observed; this cleavage was markedly reduced in CAPN2 knockdown cells (Figure 4F). Moreover, the elevation of the C-terminal fragment of filamin A induced by CoCl₂ treatment, mimicking a hypoxic condition, was attenuated by calpain 2 knockdown, and restored by reexpression of calpain 2 via transient transfection of a calpain 2 expression vector (Figures 4G, S7B,C). Our findings are consistent with previous studies demonstrating that the calpain-induced production of C-terminal fragments of filamin A promotes the nuclear accumulation of HIF1α. In addition, inhibition of filamin A cleavage by a calpain protease inhibitor supports this mechanism, although the isotype specificity of calpain was not specified [29]. Considering that HIF1α activation occurs under hypoxic conditions commonly observed in solid tumors [30], we examined hypoxic conditions within three-dimensional tumor spheroids to simulate the hypoxic microenvironment of tumoral tissue (Figure S8). Our analysis revealed a hypoxic inner region characterized by nuclear HIF1α localization in three-dimensional cultured tumor spheroids, which was notably decreased in CAPN2 knockdown tumor spheroids. Collectively, the downregulation of CAPN2 resulted in reduced filamin A cleavage, hindering the nuclear translocation and activity of HIF1α and leading to a decrease in TWIST1 expression and inhibition of the EMT process.

2.4 CAPN2 Knockdown Hinders Cancer Stemness and Reduces Metastatic Potential in TNBC

Tumor spheroids present both mesenchymal traits and stem cell-like properties. Cancer stemness is associated with a higher metastatic incidence, drug resistance, and poor prognosis in cancer treatment [31]. To assess the relationship between CAPN2 and cancer cell stemness, we analyzed the gene expressions of primary tumors and tumor spheroids. Compared to primary tumors, tumor spheroids showed increased levels of mesenchymal and HIF1α target genes, along with elevated CAPN2 expression (Figure S9A,B). In breast cancer, tumor spheroids are called mammospheres, and their growth is indicative of self-renewal capacity and cancer stemness [32]. In a mammosphere assay, the size of mammospheres was not increased in CAPN2 knockdown cells, while it significantly increased in control and CAPN1 knockdown cells (Figure S9C,D). Furthermore, cancer stemness markers, such as Oct4, Nanog, and Sox2, were downregulated in CAPN2 knockdown cells (Figure S9E,F). Notably, CAPN2 downregulation resulted in increased drug sensitivity (Figure S9G).

Next, we conducted an in vivo metastatic assay to determine whether the observed reduction in metastatic molecular phenotypes affected the metastasis in the in vivo model (Figures 5 and S10). A significant reduction in pulmonary metastasis was observed upon CAPN2 knockdown. The metastatic incidence was notably retarded in the CAPN2 knockdown group compared to the control group (Figure 5A–E). The mechanisms underlying the reduction in cancer stemness and metastatic potential by CAPN2 knockdown are briefly summarized in Figure 5F.

2.5 Isoform-Selective Pharmacological Inhibition of Calpain 2 Retards Metastasis in an In Vivo TNBC Metastasis Model

The results in this study strongly indicate that calpain-mediated filamin A cleavage and HIF1α activation, leading to tumor metastasis, are primarily regulated by calpain 2, rather than calpain 1 (Figure 5F). Given this isoform-specific function of calpain 2 in promoting TNBC metastasis, employing isoform-selective calpain 2 inhibitors could offer a precise and effective therapeutic approach for managing the metastatic progression of TNBC. In our earlier research, various calpain inhibitors were explored [33-38]. Among them, CNa 29 [39] exhibited significant specificity toward calpain 2, displaying greater than 20-fold stronger activity compared to calpain 1 (Figure 6A). To assess the impact of isoform-selective pharmacological inhibition of calpain 2 on tumor metastasis, we evaluated the effects of CNa 29 in TNBC cells. Without altering the protein level of calpain 2 (Figure S11A), CNa 29 retarded cancer cell growth in both 2D and 3D culture systems (Figure 6B,C). MDA-MB-231 cells were treated with CNa 29 to determine whether pharmacological inhibition of calpain 2 by CNa 29 prevents filamin A cleavage-induced HIF1α nuclear translocation, TWIST1 expression, EMT, cancer stemness, and metastasis. CNa 29 significantly reduced filamin A cleavage, vimentin expression, TWIST1 transcription, and the expressions of genes related to cancer stemness and increased the efficacy of doxorubicin, a chemotherapy drug commonly used for TNBC in clinical settings (Figures 6D,E and S11B–D). Furthermore, CNa 29 exhibited a significant reduction in lung colonization in an in vivo metastasis model (Figure 6F,G) without loss of body weight (Figure S12). Overall, selectively inhibiting calpain 2 with CNa 29 is a potent strategy for TNBC therapy (Figure S13).

4.1 Bioinformatic Analysis

For survival analysis, the Kaplan–Meier plotter (http://kmplot.com/breast/) was used [69]. CAPN2 or CAPN1 gene (probe set 208683_at or 200752_s_at) was analyzed, and results were displayed as a Kaplan–Meier survival plot. CAPN2 gene expression was analyzed using univariate Cox proportional analysis and correlation with mesenchymal genes using b-GenExMiner [70, 71] and TIMER [72]. Gene expression assessment in cancer cell lines was conducted using Cancer Cell Line Encyclopedia (CCLE) microarray data (GSE36133) [73]. GSEA was investigated using GSE103091, a dataset of gene expression in TNBC patients [74, 75].

4.2 Cell Culture

TNBC cell lines (MDA-MB-231, MDA-MB-436, and Hs578T) were cultured at 37°C in a humidified atmosphere with 5% CO₂. MDA-MB-231 cells were cultured in RPMI (Welgene, Korea) containing 10% fetal bovine serum (FBS, Hyclone, USA), while MDA-MB-436 and Hs578T cells were cultured in DMEM (Welgene, Korea) containing 10% FBS. Cell lines were obtained from the Korean Cell Line Bank. The MDA-MB-231-luc cell line was used in in vivo metastasis studies. Luciferase activity was tested before cell injection into the tail vein of in vivo models.

4.3 Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR)

Total RNA was extracted using TRIzol reagent (Invitrogen, USA). Reverse transcription of mRNAs was performed using the PrimeScript RT Reagent Kit (Takara, USA). qRT-PCR was performed using the SensiFAST SYBR No-ROX Kit (Bioline, Korea). GAPDH and U6 were used as internal controls, and the relative mRNA expression was detected using the CFX96 RT-PCR detection system (Bio-Rad, Korea) at the EWHA drug development research core center and the 2^−ΔΔCT method. Primer sequences used in this study are listed in Table S1.

4.4 Western Blot

Proteins were extracted using RIPA buffer and measured using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, USA). Proteins were isolated using 10%–15% SDS-PAGE and subsequently transferred to PVDF membranes (PALL Corp., USA). Membranes were blocked with 5% skimmed milk or BSA and incubated overnight with primary antibodies. The membranes were then washed and incubated with HRP-labeled secondary antibodies for 60 min at room temperature. The immune response bands were visualized using enhanced chemiluminescence (Amersham ECL Prime, GE Healthcare, USA; ChemiDoc MP, Bio-RAD, USA). All antibody information is shown in Table S2.

4.5 Immunofluorescence

Cells were seeded in eight-well chamber slides (SPL, Korea). After reaching 80% confluency, cells were washed with PBS, fixed with paraformaldehyde, and blocked with blocking solution containing 5% Blocking One-P (Nacalai Tesque, Japan) and 0.1% Triton-X in PBS. Primary antibodies were applied overnight at 4°C, followed by secondary antibodies for 1 h at room temperature. DAPI was used for nuclei staining. The cells were then washed three times, treated with mounting solution (Dako, Agilent Pathology Solutions, USA), and covered with a cover glass. Images were obtained using an apotome laser-scanning microscope at the EWHA drug development research core center and were analyzed with Zen Pro software.

4.6 Chromatin Immunoprecipitation (ChIP) Assay

Cells were seeded in a 150 mm cell culture plate (Nunc, USA) and cultured until reaching approximately 80% confluency. To assess HIF1α binding to the HRE of the TWIST1 promoter, cells were cultured in serum-free medium for 21 h, followed by an additional incubation for 3 h in serum-free medium containing 50 µM CoCl₂ prior to the ChIP assay. The ChIP assay was conducted using the Pierce Agarose ChIP Kit (Thermo Fisher Scientific, USA) according to the manufacturer's instructions. The HIF1α-bound HRE of TWIST was evaluated using qRT-PCR.

4.7 In Vivo Metastasis Study

Five-week-old female BALB/c nude mice were procured from Orient Bio (Korea). Animal handling was conducted in accordance with ethical guidelines approved by the Animal Experiment Ethics Committee of Ewha Womans University, adhering to relevant regulations (IACUC20-032, IACUC23-066). Throughout the study, mice had unrestricted access to food and water and were maintained under constant temperature and humidity with a 12 h light/dark cycle.

MDA-MB-231-luc cells (2 × 10⁵) were injected into the tail vein of the mice. Metastatic incidence and body weight were evaluated every 2–3 days. Metastatic assessment was conducted using the in vivo imaging system (IVIS) Lumina Series III (Perkin Elmer, USA) at the EWHA drug development research core center. Metastasis was measured after luciferin injection. The intensity of luminescence was quantified in the same area of the region of interest (ROI) with 60 s exposure.

To evaluate anti-metastatic activity, CNa 29 was administered intraperitoneally at doses of 15 or 30 mg/kg daily, starting the day after cell injection. In the experiment assessing metastatic inhibition by CNa 29, the body weight of each mouse was measured daily.

4.8 Statistical Analysis

Statistical analysis was performed using GraphPad Prism 10.0.0 software (GraphPad Software, USA). Data are presented as mean ± standard error of the mean (SEM). The two-tailed Student's t-test was used for comparisons between two groups, and one-way or two-way ANOVA followed by a multiple comparison test. A p value less than 0.05 (p < 0.05) was considered statistically significant.